Dadidadida

Ex 4.32.1 Power Sets

2012-01-27T21:31:00.000-08:00

Is it true that for all sets A and B, Ƥ(A ∩ B) = Ƥ(A) ∩ Ƥ(B) ?

I think so, and here is the proof:

  Let X ∈ Ƥ(A ∩ B)
  We will show that X ∈ Ƥ(A) ∩ Ƥ(B)

    From X ∈ Ƥ(A ∩ B), we get X ⊆ A ∩ B i.e., X ⊆ A and X ⊆ B
    It follows that X ∈ Ƥ(A) and X ∈ Ƥ(B), i.e., X ∈ Ƥ(A) ∩ Ƥ(B)
    Thus Ƥ(A ∩ B) ⇒ Ƥ(A) ∩ Ƥ(B)

  Let Y ∈ Ƥ(A) ∩ Ƥ(B)
  We will show that Y ∈ Ƥ(A ∩ B)

    From Y ∈ Ƥ(A) ∩ Ƥ(B), we get Y ∈ Ƥ(A) and Y ∈ Ƥ(B), i.e., Y ⊆ A and Y ⊆ B
    Let m ∈ Y.  Then m ∈ A and m ∈ B, i.e., m ∈ A ∩ B
    So Y ⊆ A ∩ B
    It follows that Y ∈ Ƥ(A ∩ B)
    Thus Ƥ(A) ∩ Ƥ(B) ⇒ Ƥ(A ∩ B)

Proof Ex 4.17.2 Intersection

2012-01-22T20:57:00.000-08:00

Show that A ∩ B = A - (A - B)

Proof:

    A - (A - B)
  = x ∈ A ∧ x ∉ (A - B)
  = x ∈ A ∧ ¬(x ∈ A ∧ x ∉ B)
  = x ∈ A ∧ (x ∉ A ∨ x ∈ B)
  = (x ∈ A ∧ x ∉ A) ∨ (x ∈ A ∧ x ∈ B)
  = x ∈ A ∧ x ∈ B

  ⇔ A ∩ B

Proof Ex 4.17.1 Subset

2012-01-22T20:29:00.000-08:00

Show that A ⊄ B ⇔ A - B ≠ ∅

Proof:

    A ⊄ B 
  = ¬∀x(x ∈ A ⇒ x ∈ B)
  = ¬∀x(x ∉ A ∨ x ∈ B)
  = ∃x(¬(x ∉ A ∨ x ∈ B))
  = ∃x(x ∈ A ∧ x ∉ B)

  ⇔ A - B ≠ ∅

Proof Ex 4.7 Set of ordinary sets

2012-01-22T10:35:00.000-08:00

Assume that A is a set of sets.  Show that {x ∈ A|x ∉ x} ∉ A.

Proof:

  Let P(x) = x ∈ A ∧ x ∉ x, and F = {x|P(x)}
  We will show that F ∉ A

    Assume the contrary, F ∈ A
    There are only two possibilities: either F ∉ F or F ∈ F

    Suppose F ∉ F
    From F ∈ A and F ∉ F, we get P(F)
    From F = {x|P(x)}, it follows F ∈ F.  A contradiction.

    Suppose F ∈ F
    From F = {x|P(x)}, we get P(F)
    From P(F) = F ∈ A ∧ F ∉ F, it follows F ∉ F.  A contradiction.

  Thus F ∉ A, which means {x ∈ A|x ∉ x} ∉ A

Proof Ex 3.43 Prime 11

2012-01-21T12:47:00.000-08:00

Prove that 11 is the only prime of the form p^2 + 2, with p prime

Proof:

  Let n = p^2 + 2
  If n is 11, p is 3
  If n is not 11, p must be either of the form 3a+1 or 3a+2, with a ∈ ℕ

  Suppose p = 3a+1
  n = p^2 + 2 = (3a+1)^2 + 2 = 9a² + 6a + 3

  Suppose p = 3a+2
  n = p^2 + 2 = (3a+2)^2 + 2 = 9a² + 12a + 6

  In both cases, n is divisible by 3 and therefore not prime
  Thus 11 is the only prime of the form p^2+2, with p prime

Proof Ex 3.36 Mersenne Prime

2012-01-17T21:31:00.000-08:00

Show that if n is composite, 2ⁿ - 1 is composite too.

(Hint: Asume that n = ab and prove that xy = 2ⁿ - 1 for the numbers x = 2^b- 1 and y = 1 + 2^b + 2^2b + ... + 2^(a-1)b)

Solution:

  Assume n = ab where a, b ∈ ℕ
  Let x = 2^b - 1 and y = 1 + 2^b + 2^2b + ... + 2^(a-1)b

  xy = (2^b - 1)(1 + 2^b + 2^2b + ... + 2^(a-1)b)
     =        2^b + 2^2b + ... + 2^(a-1)b + 2^(a-1)b+b
       - (1 + 2^b + 2^2b + ... + 2^(a-1)b)
     = 2^ab - 1
     = 2ⁿ - 1

  This shows that if n is composite, 2ⁿ - 1 is also composite since we can always find integers x and y of the form above.

Proof Ex 3.34 Prime numbers

2012-01-16T23:02:00.000-08:00

Let A = {4n + 3 | n ∈ ℕ}.  Show that A contains infinitely many prime numbers.

(Hint: any prime > 2 is odd, hence of the form 4n + 1 or 4n + 3.)

Solution:

  Assume there are only finitely many primes of the form 4n + 3, say p₁,...,p_m
  
    Consider the number N = 4p₁...p_m - 1 = 4(p₁...p_m - 1) + 3
    We will show that N must have a factor p of the form 4n + 3
  
      Suppose N does not have a factor of the form 4n + 3
      Note the fact that (4a + 1)(4b + 1) is of the form 4c + 1
      If all factors of N is of the form 4n + 1, N will be of the form 4n + 1, but N is not
      Thus N must have a factor of the form 4n + 3
  
    Based on the assumption, it follows p is a member of {p₁,...,p_m} and p divides N
    But dividing N by p would result in the form 4c - 1/p which is a fraction (where c ∈ ℕ)
    So p cannot divide N.  A contradiction.
  
  Thus there are infinitely many primes of the form 4n + 3.

Proof Ex 2.19 Equivalence and logical validity

2012-01-09T08:27:00.000-08:00

Prove that

  ϕ ≡ ψ iff ϕ ⇔ ψ

Proof:

  The above statement can be expanded into a conjunction of two sub-statements

    ((ϕ ≡ ψ) ⇒ (ϕ ⇔ ψ)) ∧
    ((ϕ ⇔ ψ) ⇒ (ϕ ≡ ψ))

  which can be further expanded into

    ((ϕ ≡ ψ) ⇒ ((ϕ ⇒ ψ) ∧ (ψ ⇒ ϕ))) ∧
    (((ϕ ⇒ ψ) ∧ (ψ ⇒ ϕ)) ⇒ (ϕ ≡ ψ))

  1) Given: ϕ ≡ ψ
     To be proven: (ϕ ⇒ ψ) ∧ (ψ ⇒ ϕ)

     Suppose ϕ and ψ
     It follows (ϕ ⇒ ψ) ∧ (ψ ⇒ ϕ)

  2) Given: (ϕ ⇒ ψ) ∧ (ψ ⇒ ϕ)
     To be proven: ϕ ≡ ψ

     Suppose ϕ, then ψ
     Suppose ¬ϕ, then ¬ψ

     Thus ϕ ≡ ψ

Thus ϕ ≡ ψ iff ϕ ⇔ ψ

Proof Ex 3.26

2012-01-07T15:05:00.000-08:00

Given

  ∀x∃y(xRy),
  ∀x∀y(xRy => yRx), 
  ∀x∀y∀z(xRy ∧ yRz => xRz) 

Prove that

  ∀x(xRx)

Proof:

  Let a be arbitrary
  From ∀x∃y(xRy), it follows ∃y(aRy)
  Let b such that aRb
  From ∀x∀y(xRy => yRx), it follows aRb => bRa
  From ∀x∀y∀z(xRy ∧ yRz => xRz), it follows aRb ∧ bRa => aRa

Thus ∀x(xRx)

Proof Ex 3.28

2012-01-07T13:20:00.000-08:00

Given

  ∀y∃z∀xP(x,y,z)

Prove that

  ∀x∀y∃zP(x,y,z)

Proof:

  Let a,b be arbitrary
  We need to show ∃zP(a,b,z)
  Given ∀y∃z∀xP(x,y,z), it follows ∃z∀xP(x,b,z)
  Let c be such that ∀xP(x,b,c)
  So P(a,b,c)

Thus ∀x∀y∃zP(x,y,z)

Proof Ex 3.27.3

2012-01-03T08:55:00.000-08:00

Given
∀x∀y(xRy ∧ x≠y => ¬yRx)
Prove that∀x∀y(xRy ∧ yRx => x=y)
Proof:

  Suppose a and b are arbitrary objects, it follows (by definition)aRb ∧ a≠b => ¬bRa
  Suppose: aRb ∧ bRa
  To be proven: a=b

  Proof:
    Suppose: a≠b
    To be proven: ⊥ (ie contradiction)

    Proof:
      From aRb ∧ a≠b => ¬bRa, and aRb ∧ bRa,
      it follows ¬bRa ∧ bRa ≡ ⊥

    Thus a=b

  Thus ∀x∀y(xRy ∧ yRx => x=y)

Yet another Luhn check and digit generation

2011-11-24T16:15:00.000-08:00

But this time it's faster:

    /**
     * @param s non-empty string that contains only digits
     * @return true only if the given digits pass the luhn check
     */
    boolean isLuhn(String s) {
        char[] ca = s.toCharArray();
        int sum = ca[ca.length-1] - '0';
        
        for (int i=ca.length-2; i >= 0; i-=2) {
            int x2 = (ca[i] - '0') << 1;
            sum += x2 < 10 ? x2 : 1 + (x2 - 10);
        }
        for (int i=ca.length-3; i >= 0; i-=2)
            sum += ca[i] - '0';
        return sum % 10 == 0;
    }

    /**
     * @param s non-empty string that contains only digits
     * @return the generated Luhn digit
     */
    int calculateLuhnDigit(String s) {
        char[] ca = s.toCharArray();
        int sum = 0;
        
        for (int i=ca.length-1; i >= 0; i-=2) {
            int x2 = (ca[i] - '0') << 1;
            sum += x2 < 10 ? x2 : 1 + (x2 - 10);
        }
        for (int i=ca.length-2; i >= 0; i-=2)
            sum += ca[i] - '0';
        int mod = sum % 10;
        return mod == 0 ? 0 : 10 - mod;
    }

Java 7 behaves differently on InetAddress.getHostName()

2011-11-19T20:39:00.000-08:00

On my Mac, if I execute the following code in Java 6:

System.out.println(InetAddress.getLocalHost().getHostName());

I got:

5855pns93223

But when run in Java 7, I got:

5855pns93223.nznmba.com

This causes problem with some existing Java mail code, with exception like:

Caused by: javax.mail.MessagingException: 501 5.0.0 HELO requires domain address
        at com.sun.mail.smtp.SMTPTransport.issueCommand(SMTPTransport.java:1363)
        at com.sun.mail.smtp.SMTPTransport.helo(SMTPTransport.java:838)
        at com.sun.mail.smtp.SMTPTransport.protocolConnect(SMTPTransport.java:375)
        at javax.mail.Service.connect(Service.java:275)
        at javax.mail.Service.connect(Service.java:156)
        at javax.mail.Service.connect(Service.java:105)
        at javax.mail.Transport.send0(Transport.java:168)
        at javax.mail.Transport.send(Transport.java:98)

This isn't good for unit testing when I need to run them locally.

Solution ?

Add two entries to my /etc/hosts:

127.0.0.1       5855pns93223
127.0.0.1       5855pns93223.nznmba.com

A hack, I know, but a quick fix at least for now :)

Scriptfu comparing memcached hit counts

2011-11-18T14:38:00.000-08:00

If we have a text file "hosts" that contains all the host names in a memcached fleet, we can readily capture a snapshot of the memcached's "get_hits" counts of the entire fleet to a file "hits", like so:

for i in `cat hosts`
do
  echo -ne 'stats\r\n' | nc -i1 $i:11211 | grep get_hits | cut -d ' ' -f3 | sed 's/[^0-9]//'
done | tee hits

(Do you see why we need to do the sed at the end ?) Anyhow, here is the fun part:

Given we have captured two files "hits.1" and "hits.2" at T1 and T2, how can we show the increments of the "get_hits" count of every host ? With a one-liner, that is :)

Solution:

paste -d '-' hits.2 hits.1 | xargs -I {} echo {} | bc -iq | paste hosts -

Neat, huh ?

How to compile with Libevent on Mac ?

2011-11-12T11:07:00.002-08:00

Assuming Libevent has been installed on Mac, and now try to compile something that uses it such as the toy low-level ROT13 server with Libevent:

$ gcc libeventBasedRot13Server.c

would failed with something like:

Undefined symbols:
  "_evutil_make_socket_nonblocking", referenced from:
      _do_accept in cc2B6M02.o
      _run in cc2B6M02.o
  "_event_new", referenced from:
      _alloc_fd_state in cc2B6M02.o
      _alloc_fd_state in cc2B6M02.o
      _run in cc2B6M02.o
  "_event_add", referenced from:
      _do_read in cc2B6M02.o
      _do_accept in cc2B6M02.o
      _run in cc2B6M02.o
  "_event_base_dispatch", referenced from:
      _run in cc2B6M02.o
  "_event_free", referenced from:
      _alloc_fd_state in cc2B6M02.o
      _free_fd_state in cc2B6M02.o
      _free_fd_state in cc2B6M02.o
  "_event_base_new", referenced from:
      _run in cc2B6M02.o
  "_event_del", referenced from:
      _do_write in cc2B6M02.o
ld: symbol(s) not found
collect2: ld returned 1 exit status

Solution:

$ gcc libeventBasedRot13Server.c /usr/local/lib/libevent.a

How to turn off class file verification in junit ant task ?

2011-11-06T09:36:00.000-08:00

Example:

    <junit printsummary="yes" haltonfailure="yes" fork="true">
      <jvmarg value="-noverify" />
      ...
    </junit>

(Why ? This was necessary to get Cobertura to run in peace with Java 7)

Scriptfu counting opened file descriptors

2011-10-27T13:13:00.000-07:00

How to count the number of opened file descriptors of all running JVM's on a Linux host ?

for ((;;))
do
  declare sum=0
  for i in `pgrep -lf jdk | awk '{print $1}'`
  do
    ((sum += `sudo /usr/sbin/lsof -p $i | wc -l`))
  done
  echo "`date` java: $sum"
  sleep 5
done | tee fd-java.log

Or a faster alternative:

for ((;;))
do
  for i in `pgrep -lf jdk | awk '{print $1}'`
  do
    s="$s$i,"
  done
  echo "`date` java: `sudo /usr/sbin/lsof -p $s | wc -l`"
  sleep 5
done | tee fd-java.log

What if you need a break down per JVM ?

for ((;;))
do
  echo -n "`date` "
  for i in `pgrep -lf jdk | awk '{print $1}'`
  do
    echo -n "[$i]:"
    sudo /usr/sbin/lsof -p $i | wc -l | tr '\n' ','
    echo -n " "
  done
  echo ""
  sleep 5
done | tee fd-jvms.log

How to install the Haskell bundle for TextMate ?

2011-10-18T16:11:00.000-07:00

Download the bundle eg

https://github.com/textmate/haskell.tmbundle/tarball/master

Uncompress the bundle eg

tar xzf textmate-haskell.tmbundle-5d41f32.tgz

rename the directory to Haskell.tmbundle eg

mv textmate-haskell.tmbundle-5d41f32 Haskell.tmbundle

mv Haskell.tmbundle /Library/Application\ Support/TextMate/Bundles/

Restart TextMate

Cryptarithm: Math - Is = Fun

2011-10-18T10:11:00.000-07:00

  MATH
-   IS
------
   FUN

Given each alphabet is a digit, what are the possible solutions ?

(Don't scroll down if you want to give it a shot)

Haskell in GHCI:

[(m,a,t,h,i,s,f,u,n)|m <- [0..9], a <- [0..9], t <- [0..9], h <- [0..9], i <- [0..9], s <- [0..9], f <- [0..9], u <- [0..9], n <- [0..9], (m*1000+a*100+t*10+h) - (i*10+s) == f*100+u*10+n, not (elem m [a,t,h,i,s,f,u,n]), not (elem a [t,h,i,s,f,u,n]), not (elem t [h,i,s,f,u,n]), not (elem h [i,s,f,u,n]), not (elem i [s,f,u,n]), not (elem s [f,u,n]), not (elem f [u,n]), not (u==n)]

Alternatively, a much faster version (in a main.hs for example):

module Main where
main :: IO ()
main = putStrLn (show 
    [(m,a,t,h,i,s,f,u,n)
    | m <- [0..9],
      a <- remove m [0..9],
      t <- remove' [m,a] [0..9],
      h <- remove' [m,a,t] [0..9],
      i <- remove' [m,a,t,h] [0..9],
      s <- remove' [m,a,t,h,i] [0..9],
      f <- remove' [m,a,t,h,i,s] [0..9],
      u <- remove' [m,a,t,h,i,s,f] [0..9],
      n <- remove' [m,a,t,h,i,s,f,u] [0..9],
      (m*1000+a*100+t*10+h) - (i*10+s) == f*100+u*10+n
    ])
    
-- remove an element once from the list
remove :: (Eq a) => a -> [a] -> [a]
remove _ [] = []
remove a (x:xs)
    | a == x = xs
    | otherwise = [x] ++ (remove a xs)
    
-- remove each element in the removal list once from the target list
remove' ::  (Eq a) => [a] -> [a] -> [a]
remove' _ [] = []
remove' [] xs = xs
remove' (a:as) bs = remove' as (remove a bs)

360 solutions:

M,A,T,H,I,S,F,U,N
0,2,3,4,5,6,1,7,8
0,2,3,4,5,8,1,7,6
0,2,3,4,7,6,1,5,8
0,2,3,4,7,8,1,5,6
0,2,3,5,4,6,1,8,9
0,2,3,5,4,9,1,8,6
0,2,3,5,8,6,1,4,9
0,2,3,5,8,9,1,4,6
0,2,3,6,4,7,1,8,9
0,2,3,6,4,9,1,8,7
0,2,3,6,8,7,1,4,9
0,2,3,6,8,9,1,4,7
0,2,3,9,6,4,1,7,5
0,2,3,9,6,5,1,7,4
0,2,3,9,7,4,1,6,5
0,2,3,9,7,5,1,6,4
0,2,4,3,5,6,1,8,7
0,2,4,3,5,7,1,8,6
0,2,4,3,6,5,1,7,8
0,2,4,3,6,8,1,7,5
0,2,4,3,7,5,1,6,8
0,2,4,3,7,8,1,6,5
0,2,4,3,8,6,1,5,7
0,2,4,3,8,7,1,5,6
0,2,4,6,5,7,1,8,9
0,2,4,6,5,9,1,8,7
0,2,4,6,8,7,1,5,9
0,2,4,6,8,9,1,5,7
0,2,5,3,6,4,1,8,9
0,2,5,3,6,9,1,8,4
0,2,5,3,8,4,1,6,9
0,2,5,3,8,9,1,6,4
0,2,5,7,6,3,1,9,4
0,2,5,7,6,4,1,9,3
0,2,5,7,9,3,1,6,4
0,2,5,7,9,4,1,6,3
0,2,5,9,7,3,1,8,6
0,2,5,9,7,6,1,8,3
0,2,5,9,8,3,1,7,6
0,2,5,9,8,6,1,7,3
0,2,6,3,7,4,1,8,9
0,2,6,3,7,9,1,8,4
0,2,6,3,8,4,1,7,9
0,2,6,3,8,9,1,7,4
0,2,6,4,7,5,1,8,9
0,2,6,4,7,9,1,8,5
0,2,6,4,8,5,1,7,9
0,2,6,4,8,9,1,7,5
0,2,6,8,7,3,1,9,5
0,2,6,8,7,5,1,9,3
0,2,6,8,9,3,1,7,5
0,2,6,8,9,5,1,7,3
0,3,1,5,4,7,2,6,8
0,3,1,5,4,8,2,6,7
0,3,1,5,6,7,2,4,8
0,3,1,5,6,8,2,4,7
0,3,1,7,4,8,2,6,9
0,3,1,7,4,9,2,6,8
0,3,1,7,6,8,2,4,9
0,3,1,7,6,9,2,4,8
0,3,4,6,5,7,2,8,9
0,3,4,6,5,9,2,8,7
0,3,4,6,8,7,2,5,9
0,3,4,6,8,9,2,5,7
0,3,4,7,5,1,2,9,6
0,3,4,7,5,6,2,9,1
0,3,4,7,9,1,2,5,6
0,3,4,7,9,6,2,5,1
0,3,4,8,5,1,2,9,7
0,3,4,8,5,7,2,9,1
0,3,4,8,9,1,2,5,7
0,3,4,8,9,7,2,5,1
0,3,5,1,6,4,2,8,7
0,3,5,1,6,7,2,8,4
0,3,5,1,8,4,2,6,7
0,3,5,1,8,7,2,6,4
0,3,5,8,6,1,2,9,7
0,3,5,8,6,7,2,9,1
0,3,5,8,9,1,2,6,7
0,3,5,8,9,7,2,6,1
0,3,6,4,7,5,2,8,9
0,3,6,4,7,9,2,8,5
0,3,6,4,8,5,2,7,9
0,3,6,4,8,9,2,7,5
0,3,6,5,7,1,2,9,4
0,3,6,5,7,4,2,9,1
0,3,6,5,9,1,2,7,4
0,3,6,5,9,4,2,7,1
0,3,7,5,8,1,2,9,4
0,3,7,5,8,4,2,9,1
0,3,7,5,9,1,2,8,4
0,3,7,5,9,4,2,8,1
0,3,7,6,8,1,2,9,5
0,3,7,6,8,5,2,9,1
0,3,7,6,9,1,2,8,5
0,3,7,6,9,5,2,8,1
0,4,1,5,2,6,3,8,9
0,4,1,5,2,9,3,8,6
0,4,1,5,8,6,3,2,9
0,4,1,5,8,9,3,2,6
0,4,1,6,2,7,3,8,9
0,4,1,6,2,9,3,8,7
0,4,1,6,8,7,3,2,9
0,4,1,6,8,9,3,2,7
0,4,1,9,5,2,3,6,7
0,4,1,9,5,7,3,6,2
0,4,1,9,6,2,3,5,7
0,4,1,9,6,7,3,5,2
0,4,2,7,5,8,3,6,9
0,4,2,7,5,9,3,6,8
0,4,2,7,6,8,3,5,9
0,4,2,7,6,9,3,5,8
0,4,2,9,5,1,3,7,8
0,4,2,9,5,8,3,7,1
0,4,2,9,7,1,3,5,8
0,4,2,9,7,8,3,5,1
0,4,5,1,6,2,3,8,9
0,4,5,1,6,9,3,8,2
0,4,5,1,8,2,3,6,9
0,4,5,1,8,9,3,6,2
0,4,5,8,6,1,3,9,7
0,4,5,8,6,7,3,9,1
0,4,5,8,9,1,3,6,7
0,4,5,8,9,7,3,6,1
0,4,6,1,7,2,3,8,9
0,4,6,1,7,9,3,8,2
0,4,6,1,8,2,3,7,9
0,4,6,1,8,9,3,7,2
0,4,7,6,8,1,3,9,5
0,4,7,6,8,5,3,9,1
0,4,7,6,9,1,3,8,5
0,4,7,6,9,5,3,8,1
0,5,1,3,2,6,4,8,7
0,5,1,3,2,7,4,8,6
0,5,1,3,8,6,4,2,7
0,5,1,3,8,7,4,2,6
0,5,1,6,2,7,4,8,9
0,5,1,6,2,9,4,8,7
0,5,1,6,8,7,4,2,9
0,5,1,6,8,9,4,2,7
0,5,1,9,3,2,4,8,7
0,5,1,9,3,7,4,8,2
0,5,1,9,8,2,4,3,7
0,5,1,9,8,7,4,3,2
0,5,2,6,3,7,4,8,9
0,5,2,6,3,9,4,8,7
0,5,2,6,8,7,4,3,9
0,5,2,6,8,9,4,3,7
0,5,2,7,3,1,4,9,6
0,5,2,7,3,6,4,9,1
0,5,2,7,9,1,4,3,6
0,5,2,7,9,6,4,3,1
0,5,2,8,3,1,4,9,7
0,5,2,8,3,7,4,9,1
0,5,2,8,9,1,4,3,7
0,5,2,8,9,7,4,3,1
0,5,3,9,6,1,4,7,8
0,5,3,9,6,8,4,7,1
0,5,3,9,7,1,4,6,8
0,5,3,9,7,8,4,6,1
0,5,6,1,7,2,4,8,9
0,5,6,1,7,9,4,8,2
0,5,6,1,8,2,4,7,9
0,5,6,1,8,9,4,7,2
0,5,6,2,7,3,4,8,9
0,5,6,2,7,9,4,8,3
0,5,6,2,8,3,4,7,9
0,5,6,2,8,9,4,7,3
0,5,6,3,7,1,4,9,2
0,5,6,3,7,2,4,9,1
0,5,6,3,9,1,4,7,2
0,5,6,3,9,2,4,7,1
0,5,7,3,8,1,4,9,2
0,5,7,3,8,2,4,9,1
0,5,7,3,9,1,4,8,2
0,5,7,3,9,2,4,8,1
0,6,1,2,3,4,5,7,8
0,6,1,2,3,8,5,7,4
0,6,1,2,7,4,5,3,8
0,6,1,2,7,8,5,3,4
0,6,1,3,2,4,5,8,9
0,6,1,3,2,9,5,8,4
0,6,1,3,8,4,5,2,9
0,6,1,3,8,9,5,2,4
0,6,1,7,2,3,5,9,4
0,6,1,7,2,4,5,9,3
0,6,1,7,9,3,5,2,4
0,6,1,7,9,4,5,2,3
0,6,1,9,3,2,5,8,7
0,6,1,9,3,7,5,8,2
0,6,1,9,8,2,5,3,7
0,6,1,9,8,7,5,3,2
0,6,2,1,3,4,5,8,7
0,6,2,1,3,7,5,8,4
0,6,2,1,4,3,5,7,8
0,6,2,1,4,8,5,7,3
0,6,2,1,7,3,5,4,8
0,6,2,1,7,8,5,4,3
0,6,2,1,8,4,5,3,7
0,6,2,1,8,7,5,3,4
0,6,2,8,3,1,5,9,7
0,6,2,8,3,7,5,9,1
0,6,2,8,9,1,5,3,7
0,6,2,8,9,7,5,3,1
0,6,3,1,4,2,5,8,9
0,6,3,1,4,9,5,8,2
0,6,3,1,8,2,5,4,9
0,6,3,1,8,9,5,4,2
0,6,3,8,4,1,5,9,7
0,6,3,8,4,7,5,9,1
0,6,3,8,9,1,5,4,7
0,6,3,8,9,7,5,4,1
0,6,7,3,8,1,5,9,2
0,6,7,3,8,2,5,9,1
0,6,7,3,9,1,5,8,2
0,6,7,3,9,2,5,8,1
0,6,7,4,8,1,5,9,3
0,6,7,4,8,3,5,9,1
0,6,7,4,9,1,5,8,3
0,6,7,4,9,3,5,8,1
0,7,1,3,2,4,6,8,9
0,7,1,3,2,9,6,8,4
0,7,1,3,8,4,6,2,9
0,7,1,3,8,9,6,2,4
0,7,1,4,2,5,6,8,9
0,7,1,4,2,9,6,8,5
0,7,1,4,8,5,6,2,9
0,7,1,4,8,9,6,2,5
0,7,1,8,2,3,6,9,5
0,7,1,8,2,5,6,9,3
0,7,1,8,9,3,6,2,5
0,7,1,8,9,5,6,2,3
0,7,1,9,3,4,6,8,5
0,7,1,9,3,5,6,8,4
0,7,1,9,8,4,6,3,5
0,7,1,9,8,5,6,3,4
0,7,2,4,3,5,6,8,9
0,7,2,4,3,9,6,8,5
0,7,2,4,8,5,6,3,9
0,7,2,4,8,9,6,3,5
0,7,2,5,3,1,6,9,4
0,7,2,5,3,4,6,9,1
0,7,2,5,9,1,6,3,4
0,7,2,5,9,4,6,3,1
0,7,3,1,4,2,6,8,9
0,7,3,1,4,9,6,8,2
0,7,3,1,8,2,6,4,9
0,7,3,1,8,9,6,4,2
0,7,4,1,5,2,6,8,9
0,7,4,1,5,9,6,8,2
0,7,4,1,8,2,6,5,9
0,7,4,1,8,9,6,5,2
0,7,4,2,5,3,6,8,9
0,7,4,2,5,9,6,8,3
0,7,4,2,8,3,6,5,9
0,7,4,2,8,9,6,5,3
0,7,4,3,5,1,6,9,2
0,7,4,3,5,2,6,9,1
0,7,4,3,9,1,6,5,2
0,7,4,3,9,2,6,5,1
0,8,1,2,4,3,7,6,9
0,8,1,2,4,9,7,6,3
0,8,1,2,6,3,7,4,9
0,8,1,2,6,9,7,4,3
0,8,2,3,5,4,7,6,9
0,8,2,3,5,9,7,6,4
0,8,2,3,6,4,7,5,9
0,8,2,3,6,9,7,5,4
0,8,2,5,3,1,7,9,4
0,8,2,5,3,4,7,9,1
0,8,2,5,9,1,7,3,4
0,8,2,5,9,4,7,3,1
0,8,2,6,3,1,7,9,5
0,8,2,6,3,5,7,9,1
0,8,2,6,9,1,7,3,5
0,8,2,6,9,5,7,3,1
0,8,3,6,4,1,7,9,5
0,8,3,6,4,5,7,9,1
0,8,3,6,9,1,7,4,5
0,8,3,6,9,5,7,4,1
0,8,4,3,5,1,7,9,2
0,8,4,3,5,2,7,9,1
0,8,4,3,9,1,7,5,2
0,8,4,3,9,2,7,5,1
0,8,5,3,6,1,7,9,2
0,8,5,3,6,2,7,9,1
0,8,5,3,9,1,7,6,2
0,8,5,3,9,2,7,6,1
0,8,5,4,6,1,7,9,3
0,8,5,4,6,3,7,9,1
0,8,5,4,9,1,7,6,3
0,8,5,4,9,3,7,6,1
0,9,1,2,4,5,8,6,7
0,9,1,2,4,7,8,6,5
0,9,1,2,6,5,8,4,7
0,9,1,2,6,7,8,4,5
0,9,1,5,4,2,8,7,3
0,9,1,5,4,3,8,7,2
0,9,1,5,7,2,8,4,3
0,9,1,5,7,3,8,4,2
0,9,1,7,5,3,8,6,4
0,9,1,7,5,4,8,6,3
0,9,1,7,6,3,8,5,4
0,9,1,7,6,4,8,5,3
0,9,2,1,4,5,8,7,6
0,9,2,1,4,6,8,7,5
0,9,2,1,5,4,8,6,7
0,9,2,1,5,7,8,6,4
0,9,2,1,6,4,8,5,7
0,9,2,1,6,7,8,5,4
0,9,2,1,7,5,8,4,6
0,9,2,1,7,6,8,4,5
0,9,2,4,5,1,8,7,3
0,9,2,4,5,3,8,7,1
0,9,2,4,7,1,8,5,3
0,9,2,4,7,3,8,5,1
0,9,3,5,6,1,8,7,4
0,9,3,5,6,4,8,7,1
0,9,3,5,7,1,8,6,4
0,9,3,5,7,4,8,6,1
1,0,2,3,4,5,9,7,8
1,0,2,3,4,8,9,7,5
1,0,2,3,7,5,9,4,8
1,0,2,3,7,8,9,4,5
1,0,3,2,4,5,9,8,7
1,0,3,2,4,7,9,8,5
1,0,3,2,5,4,9,7,8
1,0,3,2,5,8,9,7,4
1,0,3,2,7,4,9,5,8
1,0,3,2,7,8,9,5,4
1,0,3,2,8,5,9,4,7
1,0,3,2,8,7,9,4,5
1,0,3,4,5,6,9,7,8
1,0,3,4,5,8,9,7,6
1,0,3,4,7,6,9,5,8
1,0,3,4,7,8,9,5,6
1,0,3,6,5,2,9,8,4
1,0,3,6,5,4,9,8,2
1,0,3,6,8,2,9,5,4
1,0,3,6,8,4,9,5,2
1,0,4,3,5,6,9,8,7
1,0,4,3,5,7,9,8,6
1,0,4,3,6,5,9,7,8
1,0,4,3,6,8,9,7,5
1,0,4,3,7,5,9,6,8
1,0,4,3,7,8,9,6,5
1,0,4,3,8,6,9,5,7
1,0,4,3,8,7,9,5,6
1,0,4,5,6,2,9,8,3
1,0,4,5,6,3,9,8,2
1,0,4,5,8,2,9,6,3
1,0,4,5,8,3,9,6,2
1,0,4,7,6,2,9,8,5
1,0,4,7,6,5,9,8,2
1,0,4,7,8,2,9,6,5
1,0,4,7,8,5,9,6,2
1,0,5,6,7,2,9,8,4
1,0,5,6,7,4,9,8,2
1,0,5,6,8,2,9,7,4
1,0,5,6,8,4,9,7,2

... in response to another 6th grader's home work assignment :)

Cryptarithm: Take + Home = Exam

2011-10-18T09:55:00.000-07:00

  TAKE
+ HOME
------
  EXAM

Given each alphabet is a digit, what are the possible solutions ?

(Don't scroll down if you want to give it a shot)

Haskell in GHCI:

[(t,a,k,e,h,o,m,x)|t <- [0..9], a <- [0..9], k <- [0..9], e <- [0..9], h <- [0..9], o <- [0..9], m <- [0..9], x <- [0..9], (t*1000+a*100+k*10+e) + (h*1000+o*100+m*10+e) == e*1000+x*100+a*10+m, not (elem t [a,k,e,h,o,m,x]), not (elem a [k,e,h,o,m,x]), not (elem k [e,h,o,m,x]), not (elem e [h,o,m,x]), not (elem h [o,m,x]), not (elem o [m,x]), not (m==x)]

Alternatively, a much faster version (in a main.hs for example):

module Main where
main :: IO ()
main = putStrLn (show 
    [ (t,a,k,e,h,o,m,x)
    | t <- [0..9],
      a <- remove t [0..9],
      k <- remove' [t,a] [0..9],
      e <- remove' [t,a,k] [0..9],
      h <- remove' [t,a,k,e] [0..9],
      o <- remove' [t,a,k,e,h] [0..9],
      m <- remove' [t,a,k,e,h,o] [0..9],
      x <- remove' [t,a,k,e,h,o,m] [0..9],
      (t*1000+a*100+k*10+e) + (h*1000+o*100+m*10+e) == e*1000+x*100+a*10+m
    ])
    
-- remove an element once from the list
remove :: (Eq a) => a -> [a] -> [a]
remove _ [] = []
remove a (x:xs)
    | a == x = xs
    | otherwise = [x] ++ (remove a xs)
    
-- remove each element in the removal list once from the target list
remove' ::  (Eq a) => [a] -> [a] -> [a]
remove' _ [] = []
remove' [] xs = xs
remove' (a:as) bs = remove' as (remove a bs)

124 possible solutions:

T,A,K,E,H,O,M,X
0,4,1,6,5,9,2,3
0,5,7,4,3,6,8,2
0,5,9,3,2,8,6,4
0,7,1,3,2,8,6,5
0,7,3,2,1,8,4,5
0,7,3,2,1,9,4,6
0,9,1,4,3,6,8,5
0,9,1,4,3,7,8,6
0,9,2,8,7,4,6,3
0,9,2,8,7,5,6,4
0,9,5,2,1,7,4,6
0,9,5,2,1,8,4,7
1,0,2,4,3,5,8,6
1,0,2,4,3,6,8,7
1,0,3,8,7,4,6,5
1,0,4,3,2,7,6,8
1,0,4,3,2,8,6,9
1,0,5,7,6,2,4,3
1,0,5,7,6,8,4,9
1,0,7,6,5,3,2,4
1,0,7,6,5,8,2,9
1,2,5,8,7,0,6,3
1,3,0,6,5,4,2,7
1,3,2,5,4,6,0,9
1,3,4,9,7,6,8,0
1,3,5,4,2,6,8,0
1,3,8,7,5,6,4,0
1,3,8,7,6,5,4,9
1,4,8,3,2,0,6,5
1,5,0,7,6,3,4,8
1,5,6,9,7,4,8,0
1,5,7,4,3,0,8,6
1,7,3,2,0,8,4,5
1,7,3,2,0,9,4,6
1,7,6,5,4,2,0,9
1,7,9,4,2,5,8,3
1,8,3,7,5,2,4,0
1,8,5,6,4,9,2,7
1,8,7,5,3,4,0,2
1,8,7,5,3,6,0,4
1,9,5,2,0,7,4,6
1,9,5,2,0,8,4,7
1,9,8,5,3,7,0,6
2,0,1,9,7,3,8,4
2,0,1,9,7,4,8,5
2,0,1,9,7,5,8,6
2,0,4,3,1,7,6,8
2,0,4,3,1,8,6,9
2,3,4,9,6,7,8,1
2,3,4,9,7,1,8,5
2,3,5,4,1,6,8,0
2,4,5,9,7,1,8,6
2,4,8,3,1,0,6,5
2,5,9,3,0,8,6,4
2,6,1,7,5,3,4,9
2,7,0,8,5,4,6,1
2,7,1,3,0,8,6,5
2,7,6,5,3,1,0,8
2,7,9,4,1,5,8,3
2,8,3,7,5,1,4,9
2,8,7,5,3,1,0,9
3,0,1,9,6,4,8,5
3,0,2,4,1,5,8,6
3,0,2,4,1,6,8,7
3,1,2,9,6,5,8,7
3,1,4,8,5,0,6,2
3,1,4,8,5,7,6,9
3,2,5,8,4,7,6,0
3,4,5,9,6,2,8,7
3,5,7,4,0,6,8,2
3,5,7,4,1,0,8,6
3,6,7,9,5,4,8,1
3,7,0,8,4,5,6,2
3,7,0,8,5,2,6,9
3,7,6,5,2,1,0,8
3,8,7,5,1,4,0,2
3,8,7,5,1,6,0,4
3,8,7,5,2,1,0,9
3,9,1,4,0,6,8,5
3,9,1,4,0,7,8,6
3,9,2,8,4,1,6,0
3,9,8,5,1,7,0,6
4,0,1,9,5,2,8,3
4,0,1,9,5,6,8,7
4,2,5,8,3,7,6,0
4,3,2,5,1,6,0,9
4,7,0,8,3,5,6,2
4,7,6,5,1,2,0,9
4,8,5,6,1,9,2,7
4,9,2,8,3,1,6,0
5,0,1,9,4,2,8,3
5,0,1,9,4,6,8,7
5,0,7,6,1,3,2,4
5,0,7,6,1,8,2,9
5,1,4,8,3,0,6,2
5,1,4,8,3,7,6,9
5,3,0,6,1,4,2,7
5,3,8,7,1,6,4,0
5,4,1,6,0,9,2,3
5,6,1,7,2,3,4,9
5,6,7,9,3,4,8,1
5,7,0,8,2,4,6,1
5,7,0,8,3,2,6,9
5,8,3,7,1,2,4,0
5,8,3,7,2,1,4,9
6,0,1,9,3,4,8,5
6,0,5,7,1,2,4,3
6,0,5,7,1,8,4,9
6,1,2,9,3,5,8,7
6,3,4,9,2,7,8,1
6,3,8,7,1,5,4,9
6,4,5,9,3,2,8,7
6,5,0,7,1,3,4,8
7,0,1,9,2,3,8,4
7,0,1,9,2,4,8,5
7,0,1,9,2,5,8,6
7,0,3,8,1,4,6,5
7,2,5,8,1,0,6,3
7,3,4,9,1,6,8,0
7,3,4,9,2,1,8,5
7,4,5,9,2,1,8,6
7,5,6,9,1,4,8,0
7,9,2,8,0,4,6,3
7,9,2,8,0,5,6,4

... in response to a 6th grader's home work assignment :)

List Comprehension in Haskell

2011-10-06T22:41:00.000-07:00

To use Haskell to find a right angle triangle that has:

all three sides as integers
each side less than or equal to 10
a perimeter of 24

In ghci, type:

[(a,b,c) | c <- [1..10], b <- [1..c], a <- [1..b], a+b+c == 24, a^2+b^2 == c^2]

Output:

[(6,8,10)]

Pretty amazing :)

How to combine multiple PDF files into one in Mac OS X ?

2011-06-09T15:55:00.001-07:00


python '/System/Library/Automator/Combine PDF Pages.action/Contents/Resources/join.py' -o 'output.pdf' *.pdf

How to recover /home from Snow Leopard ?

2010-09-13T00:03:00.000-07:00

Rename the following line in /etc/auto_master :

  /home auto_home -nobrowse

To:

  /home_auto auto_home -nobrowse

More details

Using int or long as bit array

2010-08-28T20:50:00.000-07:00

Ever consider making use of a primitive int or long as a bit array in Java ? An int, for example, would provide a capacity of 32 bits. But how would we go about setting and testing the nth bit efficiently ?

Possible solutions:

(Don't peek if you want to give it a try!)

    int setBitOn(int in, int pos) { return in | 1 << pos; }
    int setBitOff(int in, int pos) { return in & (-1 ^ (1 << pos)); }
    boolean isBitOn(int in, int pos) { return (in & (1 << pos)) != 0; }
    boolean isBitOff(int in, int pos) { return (in & (1 << pos)) == 0; }

Similarly for long,

    long setBitOn(long in, int pos) { return in | 1L << pos; }
    long setBitOff(long in, int pos) { return in & (-1L ^ (1L << pos)); }
    boolean isBitOn(long in, int pos) { return (in & (1L << pos)) != 0; }
    boolean isBitOff(long in, int pos) { return (in & (1L << pos)) == 0; }

Any faster, better or simpler way to do this ?

File Diff %

2010-03-08T01:51:00.000-08:00

Given two text files v1.txt and v2.txt, how can we easily tell the percentage of total changes (including lines added, changed, and deleted) from v1.txt to v2.txt ?

Additional Puzzle

2010-03-07T20:38:00.000-08:00

In this little piece of addition, every digit from 0-9 is uniquely represented by a letter of the alphabet. It may be assumed that T is not zero. Can you work out the values of the various letters ?

    TWENTY
    TWENTY
+   THIRTY
----------
   SEVENTY

What is the next line ?

2010-02-02T16:35:00.000-08:00

3 4 1 2 2
4 5 2 0 0
5 6 3 0 0
6 7 4 2 8
7 8 5 6 3 0

SHA-1 and MD5: Mac vs Java

2009-11-03T09:40:00.000-08:00

echo -n "" | openssl dgst -sha1
=> da39a3ee5e6b4b0d3255bfef95601890afd80709

echo -n "" | md5
=> d41d8cd98f00b204e9800998ecf8427e

public class FooCrypt {
    public static void main(String[] args) throws Exception {
        MessageDigest sha1 = MessageDigest.getInstance("SHA-1");
        System.out.println("sha-1: " + bytesToHex(sha1.digest("".getBytes("UTF-8"))));
        MessageDigest md5 = MessageDigest.getInstance("MD5");
        System.out.println("md5: " + bytesToHex(md5.digest("".getBytes("UTF-8"))));
    }

    public static String bytesToHex(byte[] bs) {
        StringBuilder ret = new StringBuilder(bs.length);
        for (int i = 0; i < bs.length; i++) {
            String hex = Integer.toHexString(0x0100 + (bs[i] & 0x00FF)).substring(1);
            ret.append((hex.length() < 2 ? "0" : "")).append(hex);
        }
        return ret.toString();
    }
}

=> 
sha-1: da39a3ee5e6b4b0d3255bfef95601890afd80709
md5: d41d8cd98f00b204e9800998ecf8427e

What comes next ?

2009-10-26T18:15:00.000-07:00

3
1 3
1 1 1 3
3 1 1 3
1 3 2 1 1 3
1 1 1 3 1 2 2 1 1 3
3 1 1 3 1 1 2 2 2 1 1 3
1 3 2 1 1 3 2 1 3 2 2 1 1 3

Experimenting with Oracle DBMS_LOCK

2009-10-25T16:56:00.000-07:00

First, grant privilege to play with it. For example,

Login as user SYS with SYSDBA role, then

  grant all on dbms_lock to public;

How to try it out in SQL Developer ?

set serveroutput on;

declare
  v_result integer;
begin
  v_result := DBMS_LOCK.REQUEST(123, 6, 0, true);
  
  dbms_output.put_line(
           case 
              when v_result=0 then 'Success'
              when v_result=1 then 'Timeout'
              when v_result=2 then 'Deadlock'
              when v_result=3 then 'Parameter Error'
              when v_result=4 then 'Already owned'
              when v_result=5 then 'Illegal Lock Handle'
           end);
end;

How to do it in JDBC ?

  Connection conn = ...
  CallableStatement stmt = conn.prepareCall("{call ? := DBMS_LOCK.REQUEST(?, 6, 0, true)}");
  stmt.registerOutParameter(1, Types.INTEGER);
  stmt.setBigDecimal(2, new BigDecimal(123));
  stmt.execute();
  int result = stmt.getInt(1);
  System.out.println("result: " + result);

S3/SimpleDB Limitations

2009-09-07T18:10:00.000-07:00

S3 Limits - max 100 buckets per account, unlimited number of objects per bucket with object size < 5GB

SimpleDB Limits - max 10 GB per domain, max 100 domains per account, max 1KB attribute value, max 256 attributes per item

Covariant Enum Idiom

2009-04-25T10:54:00.000-07:00

Say we want to have a Java Enum that can return different types from the same method by different constants, how could we go about writing it ? How about something like:

/**
 * An enum that can polymorphically return different types of object.
 */
public enum CovariantEnum {
    A { 
        @Override public String getValue() { return "CovariantEnum.A"; } 
    },
    B {
        @Override public Integer getValue() { return 1; }
    },
    // etc.
    ;
    public abstract <T> T getValue();
}

This would actually compile, except there are two problems. First, there are compilation warnings about "return type requires unchecked conversion". Second, it actually doesn't quite work as expected. Client code such as:

// This doesn't work :(
String stringValue = CovariantEnum.A.getValue();

would result in a type mismatch compilation error. Explicit casting the returned value to the target type, such as String in the above example, would work, but that defeats the purpose of building CovariantEnum in the first place.

So how can we get around these limitations ?

Well, first, Java disallows enum declaration to have type parameters. In other words, it is a syntax error to write:

public enum CovariantEnum<T> {
...

But what if we emulate a enum with an abstract class ? This may lead to something like:

/**
 * Basically a "CovariantEnum" that actually works and does not cause any compilation warnings.
 */
public abstract class TypeSafeCovariantEnum<T> {
    public static final TypeSafeCovariantEnum<String> A = new TypeSafeCovariantEnum<String>() { 
        @Override public String getValue() { return "TypeSafeCovariantEnum.A"; } 
    };
    public static final TypeSafeCovariantEnum<Integer> B = new TypeSafeCovariantEnum<Integer>() { 
        @Override public Integer getValue() { return 1; }
    };
    
    private TypeSafeCovariantEnum() {}
    public abstract T getValue();
}

Now compilation has zero warnings, and a client can enjoy using the "literal instances" in TypeSafeCovariantEnum as if it's a covariant enum without any need for explicit casting:

// This works!
String stringValue = TypeSafeCovariantEnum.A.getValue();
Integer integerValue = TypeSafeCovariantEnum.B.getValue();

Of course this is not an actual enum, and it involves a little complexity inside; but it provides 100% type safety, and simplicity to the clients outside :-)

Local vs Any address in Cajo

2008-11-30T00:24:00.000-08:00

Looking into the cajo source code, I see the default server side host address used by cajo is the local host address. For example,

host = InetAddress.getLocalHost().getHostAddress();

This seems to work in most cases/platforms, except it breaks with the new Mac that I have. How does it break ? Well the server side binding works with the default (local) address, but the client would fail to connect to 127.0.0.1 or localhost. To make the client connect, I needed to figure out the specific local address (like 192.168.0.3) used for the server side binding, and explicitly specify that for the client.

When I tried to see if this was also the case outside cajo using the plain old ServerSocket(int), surprisingly it worked regardless of whether I specified locahost, 127.0.0.1, or 192.168.0.3.

Digging more into the JDK source code, ServerSocket uses an "any address" rather than a "local address" as the default:

InetAddress.anyLocalAddress();

Sadly, this static method is package private.

Anyhow, since the "any address" seems to work more so than the "local address", it would be nice if cajo can change the default server side address to the "any address", which would make cajo "just work" across more platforms.

Warming up a JVM

2008-11-22T07:51:00.000-08:00

If you knew the actual set of classes that will be lazily loaded into the JVM for execution, wouldn't it be nice if these classes can be eagerly loaded into the JVM upon start-up ? This particularly matters if you are writing web services, where the latency of the first few requests will be significantly impacted.

Pre-loading classes is easy. But how can we find out the precise set of classes to be preloaded ? It turns out there exists a very nice JVM option:

-verbose:class

Also found an interesting technical report: Eager Class Initialization For Java.

A little Java constructor pattern

2008-11-11T11:43:00.000-08:00

Say we have a class Item:

public class Item {
    final String name;
    final float val;
    final int size;
    
    public Item(String name, int size, float val) {
        this.name = name;
        this.val = val;
        this.size = size;
    }
}

and we'd like to initialize an Item array with some values:

Item[] items = {
    new Item("A", 3, 4),
    new Item("B", 4, 5),
    ...
};

Wouldn't it be nice if we can simplify it to:

Item[] items = {
    Item("A", 3, 4),
    Item("B", 4, 5),
    ...
};

?

Couldn't this be easily done by defining a static factory method with the same name as Item, and then use it via static import ?

public class Item {
    // ... same as above

    // A static factory method for the syntactic sugar
    public static Item Item(String name, int size, float val) {
        return new Item(name, size, val);
    }
}

Usage:

import static Item.Item;
...
Item[] items = {
    Item("A", 3, 4),
    Item("B", 4, 5),
    ...
};

Beanlib on BeanUtils wiki

2008-09-23T16:14:00.000-07:00

It made my day when I proposed to use Beanlib to solve a problem which the Jakarta Commons BeanUtils was not well suited, Beanlib ended up being included in the BeanUtils wiki as a related project :)

Thanks to Niall Pemberton for being open minded.

No more infinitely blocking socket ?

2008-09-22T12:35:00.000-07:00

One common problem Java programmer faces is the direct or indirect use of socket that could block forever. If a socket connect or read operation blocks with an infinite timeout, there is basically nothing that can interrupt the blocking thread besides closing the socket. Not Thread.interrupt(), nor even Thread.stop() would help. Gaining access to the underlying socket could, however, be difficult, considering the socket could be buried deep down inside some third party library, such as a JDBC driver.

Attending the NFJS session, Busy Java Developer's Guide to JDK Hacking by Ted Neward yesterday, a solution came to my mind when Ted demonstrated how easy it was to slip in my own classes ahead of the JDK's. Why not simply extend the existing JDK socket implementation with an additional system property to limit or default the socket timeout to some sensible values ? See an example implementation below. To make this work, simply

Set up a custom socket factory:

Socket.setSocketImplFactory(new SocketImplFactory() {
    @Override
    public SocketImpl createSocketImpl() { return new NicerSocksSocketImpl(); }
});

Slip in the implementation class and configure the socket timeout when starting up the JVM:
```
java -Xbootclasspath/p:"/path/to/custom/jdk/classes" -DSO_TIMEOUT=1000 ...
```

Now if a socket timeout is not specified by the individual application or library, the blocking operation would be limited to a maximum of 1,000ms before throwing up.

No more socket black hole sucking up resources:)

package java.net;
import java.io.IOException;

public class NicerSocksSocketImpl extends SocksSocketImpl {
    private final int defaultSoTimeout; 

    public NicerSocksSocketImpl() { this.defaultSoTimeout = defaultSoTimeout(); }

    public NicerSocksSocketImpl(String server, int port) {
        super(server, port);
        this.defaultSoTimeout = defaultSoTimeout();
    }

    public NicerSocksSocketImpl(Proxy proxy) {
        super(proxy);
        this.defaultSoTimeout = defaultSoTimeout();
    }
    
    private int defaultSoTimeout() {
        String defaultSoTimeoutStr = System.getProperty("SO_TIMEOUT");
        try {
            return Integer.parseInt(defaultSoTimeoutStr);
        } catch(RuntimeException ex) {
            return 5000;    // default to 5 seconds
        }
    }
    
    @Override
    protected void connect(SocketAddress endpoint, int timeout) throws IOException {
        super.connect(endpoint, timeout == 0 ? defaultSoTimeout : timeout);
    }

    @Override
    protected void connect(String host, int port) throws UnknownHostException, IOException {
        setTimeoutIfNecessary();
        super.connect(host, port);
    }
    
    @Override
    protected void connect(InetAddress address, int port) throws IOException {
        setTimeoutIfNecessary();
        super.connect(address, port);
    }
    
    private void setTimeoutIfNecessary() throws SocketException {
        int timeout = super.getTimeout();
        
        if (timeout == 0)
            super.setOption(SO_TIMEOUT, defaultSoTimeout);
        return;
    }
}

When is NEVER ever ?

2008-09-20T21:29:00.000-07:00

During one of today's NFJS sessions, Transaction Design Patterns by Mark Richards, the audience were asked when it would make sense to specify the transaction attribute NEVER in a real-life application.

My immediate answer to this was that if a method is known to never get involved in a transactional context, then specifying NEVER would allow the application to fail fast. But what would be the practical circumstances where a method should never get involved in a transaction, to the extent of potentially causing an exception in production ? And if there weren't really such circumstances, why not use the SUPPORTS attribute instead ? Asked Richard.

On second thought, maybe we can use the NEVER attribute a lot more often and useful than it now is. Let me explain.

Consider the transaction attribute REQUIRES_NEW, which basically creates a new transaction if one doesn't already exist. But what if there already exists a transaction ? Instead of throwing an exception, like NEVER, it would suspend the current transaction and creates a new one. However, suspending a transaction is both expensive and usually not the intended semantics, as Richard pointed out.

So wouldn't it be nice if there exists a transaction attribute

NEVER_REQUIRES_NEW

that would both guarantee there is no pre-existing transaction when a method is invoked, and then proceed to always create a new transaction ? But there is no such transaction attribute either in JEE 5, nor in Spring.

Here is what I think could achieve the NEVER_REQUIRES_NEW transaction semantics, assuming the service method name is M:

Split method M into M1 and M2, and move the implementation of M to M2;
M1 is a no-op annotated with the transaction attribute NEVER, whereas M2 is annotated with REQUIRES_NEW;
Provide a proxy service that always invokes M1 followed by M2, exposing in turn to the client only a service method M without any transactional annotation;

Now when client invokes M, the proxy would in turn invoke M1 that guarantees there is no pre-existing transaction, and then invoke M2 which would always create a new transaction, effectively resulting in NEVER_REQUIRES_NEW!

Is this yet another Transaction Design Pattern ? The use of NEVER could be forever changed :)

Visiting Every Permutation ?

2008-09-06T23:03:00.000-07:00

Given a set of objects, sometimes it's useful to find out all the permutation of these objects, and perform some action upon each permutation. The order of the permutation generation would preferably be in some natural order such as lexicographic i.e. the order they would appear if they were sorted numerically.

How would you go about writing such utility efficiently in Java ?

One possible answer:

public class Permutation<T> {
    public static interface Visitor<T> {
        /** Visits each permutation of objects. */
        public void visit(T[] entry);
    }
    private final T[] objects;
    private T tmp;
    private final Visitor<T> visitor;
    
    /**
     * @param objects set of n objects
     * @param visitor used to visit each permutation
     */
    public Permutation(T[] objects, Visitor<T> visitor) {
        this.objects = objects.clone();
        this.visitor = visitor;
    }
    
    public void compute() { doCompute(0); }
    
    private void doCompute(final int pos) {
        if (pos == objects.length) {
            visitor.visit(objects);
            return;
        }

        doCompute(pos+1);
        
        for (int i=pos+1; i < objects.length; i++) {
            swap(pos, i);
            doCompute(pos+1);
        }
        
        if (pos > 0)
            for (int i=pos+1; i < objects.length; i++)
                swap(i-1, i);
    }
    
    private void swap(int p1, int p2) {
        tmp = objects[p1];
        objects[p1] = objects[p2];
        objects[p2] = tmp;
    }
}

Demo:

Visitor<Character> visitor = new Visitor<Character>() {
    private Set<String> set = new HashSet<String>();

    @Override
    public void visit(Character[] entry) {
        String s = Arrays.toString(entry);
        set.add(s);
        System.out.println(set.size() + ": " + s);
    }
};
String s = "abc";
Character[] ca = new Character[s.length()];

for (int i = s.length() - 1; i > -1; i--)
    ca[i] = s.charAt(i);
new Permutation<Character>(ca, visitor).compute();

Would yield:

1: [a, b, c]
2: [a, c, b]
3: [b, a, c]
4: [b, c, a]
5: [c, a, b]
6: [c, b, a]

Visiting Every Combination ?

2008-09-02T23:22:00.000-07:00

Given a set of objects, sometimes it's useful to find out all the combination of these objects of a given size, and perform some action upon each combination. For example, given a set of letters "abcde", print out all the combination of 4 letters from the set.

How would you go about writing such utility efficiently in Java ?

One possible answer:

public class Combination<T> {
    public static interface Visitor<T> {
        /** Visits each k-combination of objects. */
        public void visit(T[] entry);
    }
    private final T[] objects; // Given set of objects
    private final T[] entry;   // each combination of size k
    private final Visitor<T> visitor;

    /**
     * @param objects set of n objects
     * @param k size of each combination
     * @param visitor used to visit each combination
     */
    public Combination(T[] objects, int k, Visitor<T> visitor) {
        this.objects = objects.clone();
        entry = Arrays.copyOf(this.objects, k);
        this.visitor = visitor;
    }

    public void compute() { doCompute(0, 0); }

    private void doCompute(int pos, int entryPos) {
        if (entryPos == entry.length) {
            visitor.visit(entry);
            return;
        }
        // (entry.length - entryPos) is the remaining size
        for (int i = pos, end = objects.length - (entry.length - entryPos) + 1; i < end; i++) {
            entry[entryPos] = objects[i];
            doCompute(i + 1, entryPos + 1);
        }
    }
}

Demo:

Visitor<Character> visitor = new Visitor<Character>() {
    private Set<String> set = new HashSet<String>();

    @Override
    public void visit(Character[] entry) {
        String s = Arrays.toString(entry);
        set.add(s);
        System.out.println(set.size() + ": " + s);
    }
};
String s = "abcde";
Character[] ca = new Character[s.length()];

for (int i = s.length() - 1; i > -1; i--)
    ca[i] = s.charAt(i);

final int size = 4;
new Combination<Character>(ca, size, visitor).compute();

Would yield:

1: [a, b, c, d]
2: [a, b, c, e]
3: [a, b, d, e]
4: [a, c, d, e]
5: [b, c, d, e]

Root Cause Java Quiz

2008-08-05T12:46:00.000-07:00

Can the following code leads to an endless loop ?

    public static Throwable getRootCause(Throwable t) {
        if (t == null)
            return t;
        Throwable cause = t.getCause();
        
        if (cause == null)
            return t;
        return getRootCause(cause);
    }

Answer:

Don't try this at home ;)

        Exception ex1 = new Exception("ex1");
        Exception ex2 = new Exception("ex2");
        ex1.initCause(ex2);
        ex2.initCause(ex1);
        getRootCause(ex1);

Java Enum Puzzler

2008-07-12T17:59:00.000-07:00

What does the following print ?

public enum FooEnum {
    FOO { @Override public void bar() {} };
    public abstract void bar();
    
    public static void main(String...args) {
        System.out.println(FOO.getClass().isEnum());
    }
}

Answer:

FOO is an enum value in FooEnum. The class of FOO must therefore be an enum, and it should print "true". Right ? If you give it a try, surprisingly it will print "false". Why ?

Peeking into the JDK source code of Class.java,

public boolean isEnum() {
    // An enum must both directly extend java.lang.Enum and have
    // the ENUM bit set; classes for specialized enum constants
    // don't do the former.
    return (this.getModifiers() & ENUM) != 0 && 
            this.getSuperclass() == java.lang.Enum.class;
}

Now if you tried to print out the super class of FOO.class, it would print FooEnum.class, rather than Enum.class. The comparison of the super class of FOO.class against Enum.class would therefore fail in the second condition of the return statement.

To me, this looks like a bug either in the javac compiler, or in the implementation of Class.isEnum().

Why is it probably a bug in the javac compiler ? Well, shouldn't the super class of FOO.class be Enum.class, instead of FooEnum.class ? (Special thanks to Dhanji R. Prasanna for pointing this out.)

Why is it probably a bug in Class.isEnum() ? Well, shouldn't it recursively check all the super classes of FOO.class for Enum.class, instead of just the direct super class ? And why does it need to compare the super class with Enum.class at all, when there is already the checking for the ENUM class modifier ?

What do you think ?

Update on 7/13/2008: Please vote for bug 6710708.

Dumping a JavaBean as name/value pairs ?

2008-07-11T03:11:00.000-07:00

Sometimes it's useful to deeply dump out the properties of a complex JavaBean as name/value pairs, perhaps for debugging purposes, even when none of the toString methods of the classes involved are defined. This is especially the case when the JavaBean classes are written by a third party.

Now how would you go about doing that ?

How about using the Jarkarta commons-lang's ToStringBuilder.reflectionToString method ? It doesn't cut it, as it relies on the toString method of the individual classes involved.

How about XMLEncoder or XStream ? They work pretty nicely. However, by default XStream generates not just the public properties but all other fields as well. In both cases, the generated XML would need to be further transformed into name/value pairs.

Another way I can think of is to take advantage of the latest BeanSourceHandler SPI as a side effect of performing a deep clone via the open-source library Beanlib (beanlib-3.3.0beta18). What do I mean ? Here is an example:

Object bean = ...
// Dump out the entire JavaBean
// when none of the toString methods are defined.
BeanReplicator.newBeanReplicatable(customTransformer()).replicateBean(bean);

private static BeanTransformerSpi customTransformer() {
    BeanTransformerSpi beanTransformer = BeanTransformer.newBeanTransformer();
    return beanTransformer.initBeanSourceHandler(new BeanSourceHandler() {
        public void handleBeanSource(Object fromBean, Method readerMethod, Object propertyValue) {
            System.out.println(
                String.valueOf(fromBean.getClass().getSimpleName() + "."
                + readerMethod.getName() + "=" + propertyValue));
        }
    });
}

Little Java Quiz - File loaded from classpath

2008-07-02T10:32:00.000-07:00

Given:

a file with a given name;
the file is located in the file system;
the file can be loaded via the classpath;

How can the code figure out the physical location (ie absolute path) of the file ?

(Don't peek if you want to give it a try!)

Answer:

    String filePath = Thread.currentThread()
                            .getContextClassLoader()
                            .getResource(filename)
                            .getFile();

Tomcat 5.5 SSL Programming Puzzle

2008-04-11T00:20:00.000-07:00

Assuming SSL client side authentication is enabled in Tomcat 5.5, how can a webapp go about retrieving the underlying client's X509 certificate of the SSL socket ?

Browsing through the Tomcat source code, such information can be found in the SSL session, which unfortunately seems totally unreachable from the servlet layer.

Well, thanks to reflection, below is one such solution. Do you know of a better one ?

    /** 
     * Returns the requester's X509 SSL certificate(s) of the given request;
     * or null if no such certificate can be determined. 
     */
    static X509Certificate[] getPeerCertificates(HttpServletRequest req) {
        // rip open the tomcat specific request object to retrieve the peer SSL cert(s)
        try 
        {   // Dig out the org.apache.catalina.connector.Request object
            Field f = req.getClass().getDeclaredField("request");
            f.setAccessible(true);
            Object catalinaConnectorRequest = f.get(req);
            
            // Dig out the org.apache.coyote.Request object
            // by invoking org.apache.catalina.connector.Request.getCoyoteRequest()
            Method getCoyoteRequestMethod = catalinaConnectorRequest.getClass().getMethod("getCoyoteRequest");
            Object coyoteRequest = getCoyoteRequestMethod.invoke(catalinaConnectorRequest);

            // Dig out the proper class loader
            Class coyoteRequestClass = coyoteRequest.getClass();
            ClassLoader classLoader = coyoteRequestClass.getClassLoader();
            
            // Dig out the ActionCode class loaded by the approprate class loader
            Class actionCodeClass = classLoader.loadClass("org.apache.coyote.ActionCode");
            
            // Refer to the ActionCode.ACTION_REQ_SSL_CERTIFICATE
            Field actionReqSslCertField = actionCodeClass.getDeclaredField("ACTION_REQ_SSL_CERTIFICATE");
            Object actionReqSslCert = actionReqSslCertField.get(null);
            
            // Invoke org.apache.coyote.Request.action(ActionCode.ACTION_REQ_SSL_CERTIFICATE, null)
            Method actionMethod = coyoteRequestClass.getMethod("action", actionReqSslCert.getClass(), Object.class);
            actionMethod.invoke(coyoteRequest, actionReqSslCert, null);
            
            // Dig out the SSLSupport class loaded by the approprate class loader
            Class sslSupportClass = classLoader.loadClass("org.apache.tomcat.util.net.SSLSupport");
            
            // Refer to the SSLSupport.CERTIFICATE_KEY
            Field sslSupportCertKeyField = sslSupportClass.getDeclaredField("CERTIFICATE_KEY");
            Object sslSupportCertKey = sslSupportCertKeyField.get(null);
            
            // Invoke org.apache.coyote.Request.getAttribute(SSLSupport.CERTIFICATE_KEY)
            Method getAttributeMethod = coyoteRequest.getClass()
                                                     .getMethod("getAttribute", sslSupportCertKey.getClass());
            return (X509Certificate[])getAttributeMethod.invoke(coyoteRequest, sslSupportCertKey);
        } catch(Throwable t) {
        }
        return null;
    }

It turns out there is a much simpler way to solve this problem. According to the spec (section SRV.4.7), if there is an SSL certificate associated with the request, it must be exposed by the servlet container to the servlet programmer as an array of objects of type java.security.cert.X509Certificate and accessible via a ServletRequest attribute of javax.servlet.request.X509Certificate.

Thanks to Mark Thomas for pointing this out.

On Frugality

2008-04-03T10:00:00.000-07:00

Be contra-variant in what you take, but covariant in what you give. In other words, take less and give more.

Some JMX Quizzes

2008-02-15T15:43:00.000-08:00

Where can you find out all those Platform MBean Object Names ?
How would you connect via JMX with username and password programmatically ?

Some possible answers:

Where can you find out all those Platform MBean Object Names ?

java.lang.management.ManagementFactory

How would you connect via JMX with username and password programmatically ?

    Map<String,String[]> env = new HashMap<String,String[]>();
    env.put(JMXConnector.CREDENTIALS, new String[]{"username", "password"});
    final String url = "service:jmx:rmi://" + host + "/jndi/rmi://" + host + ":" + port + "/jmxrmi";
    JMXServiceURL jmxServiceUrl = new JMXServiceURL(url);
    JMXConnector conn = JMXConnectorFactory.connect(jmxServiceUrl, env);
    MBeanServerConnection mbsc = conn.getMBeanServerConnection();
    ...

Some Quizzes on Bits, Bytes, etc.

2008-02-13T21:47:00.000-08:00

Given a byte, how would you turn the nth bit on ?
Given a byte, how would you turn the nth bit off ?
Given a long, how would you return 8 bytes (in big-endian order) containing the two's-complement binary representation of the long ?
Given 8 bytes (in big-endian order) containing the two's-complement binary representation of a long, how would you return the long ?
Given a UUID, how would you return a BigInteger representing the numeric value of the underlying 128-bit ?

Some possible answers:

(Don't peek if you want to give it a try!)

Given a byte, how would you turn the nth bit on ?

    public static byte setBitOn(byte b, int n) {
        return (byte)(b | 1 << n);
    }

Given a byte, how would you turn the nth bit off ?

    public static byte setBitOff(byte b, int n) {
        int mask = 0xFF - (1 << n);
        return (byte)(b & mask);
    }

Given a long, how would you return 8 bytes (in big-endian order) containing the two's-complement binary representation of the long ?

    public static byte[] longTobytes(long val) {
        ByteBuffer bb = ByteBuffer.allocate(8);
        bb.putLong(val);
        return bb.array();
    }

Given 8 bytes (in big-endian order) containing the two's-complement binary representation of a long, how would you return the long ?

    public static long bytesTolong(byte[] ba) {
        ByteBuffer bb = ByteBuffer.allocate(8);
        bb.put(ba);
        bb.flip();
        return bb.getLong();
    }

Given a UUID, how would you return a BigInteger representing the numeric value of the underlying 128-bit ?

    public static BigInteger toBigInteger(UUID uuid) {
        long msb = uuid.getMostSignificantBits();
        long lsb = uuid.getLeastSignificantBits();
        ByteBuffer bb = ByteBuffer.allocate(16);
        bb.putLong(msb).putLong(lsb);
        return new BigInteger(bb.array());
    }

Tuning JVM with ICMS

2008-01-30T21:18:00.000-08:00

Some readings:

http://www.sun.com/bigadmin/content/submitted/cms_gc_logs.html
http://java.sun.com/docs/hotspot/gc5.0/gc_tuning_5.html
http://java.sun.com/docs/hotspot/gc1.4.2/example.html
http://blogs.sun.com/jonthecollector/entry/when_the_sum_of_the

http://java.sun.com/j2se/1.5.0/docs/guide/vm/gc-ergonomics.html
http://performance.netbeans.org/howto/jvmswitches/index.html
http://java.sun.com/javase/6/docs/technotes/guides/vm/cms-6.html

http://java.sun.com/performance/reference/whitepapers/tuning.html
http://blogs.sun.com/watt/resource/jvm-options-list.html

File Distribution and Replication via JGroups

2007-12-28T17:58:00.000-08:00

Here is how I used JGroups (2.6.1) to distribute and replicate files retrieved from a feed, so there is no single point of failure:

http://www.nabble.com/Re%3A-Streaming-Message-Transfer---p14510203.html

Here is a reply from Bela Ban basically confirming that the scheme works (via configuring a TCP stack):

http://www.nabble.com/Re%3A-Streaming-Message-Transfer---p14512866.html

The whole thread:

http://www.nabble.com/Streaming-Message-Transfer---td14476343.html

JGroups rocks!

Update on 13Nov08: Or does it ? After using JGroups for about a year now, it seems JGroups can be quite unreliable when the number of nodes in a cluster grow beyond half a dozen of nodes, especially if they span across a WAN. Or maybe it's just me ?

TextIterable ?

2007-10-24T15:35:00.000-07:00

Ever wonder why there doesn't exist a kind of Iterable in Java that can work with text file ? For example:

Iterable ti = new TextIterable(myfile);

for (String line: ti) {
  // do stuff with line
}
// file automatically opened and closed

Or :

TextIterable ti = new TextIterable(myfile);
int count = 0;
        
for (String line : ti) {
  if (++count > 10)
      break;
  // do stuff with line
}
ti.close();  // explicitly close it

You would think something so obvious/simple/needed that someone would have done it already. But search I may, I couldn't find any. So I ended up building one. See code below for the TextIterable and a helper class LineIterator. Note:

You don't need to explicitly open the file and you don't have to explicitly close the file. As long as you iterate through the entire file, you get the opening/closing automatically done. Yet to ensure the underlying resource is closed even in the face of exceptions, the TextIterable can be explicitly closed in a finally block.
Or, if you choose to iterate through only part of the file, it can be explicitly closed afterwards.
Each line is loaded on demand, so it would incur minimal memory footprint.
Not only does TextIterable support file, it also support resource path, and for that matter, any url!
TextIterable is threadsafe, but the iterators it produces (ie LineIterator) is not (ie intended to be used in individual threads.)
Support Encodings.
Support Fluent API for configurations.
Think File, URL, resource, rather than Stream, Reader, etc.

I think it makes the code much simpler now. What do you think ?

Special thanks to John Xiao for listening and triggering the support of (2), "lumpynose" for various suggestions, Henri Yandell for triggering the support of (6), and Dr. Heinz M. Kabutz for reminding me not to under-estimate the force ;)

/**
 * @author Hanson Char
 */
@ThreadSafe
public class TextIterable implements Iterable<String>, Closeable {
    private final URL url;
    private final List<LineIterator> openedIterators = new ArrayList<LineIterator>();
    private volatile boolean returnNullUponEof;
    private String charsetname;
    private Charset charset;
    private CharsetDecoder charsetDecoder;

    public TextIterable(File file) throws MalformedURLException {
        this(file.toURI().toURL());
    }

    public TextIterable(URL url) {
        this.url = url;
    }

    public TextIterable(String resourcePath) {
        this(
            Thread.currentThread()
                  .getContextClassLoader()
                  .getResource(resourcePath));
    }

    public LineIterator iterator() {
        LineIterator ret;
        final String charsetname;
        final Charset charset;
        final CharsetDecoder charsetDecoder;
        
        synchronized(this) {
            charsetname = this.charsetname;
            charset = this.charset;
            charsetDecoder = this.charsetDecoder;
        }
        try {
            if (charsetDecoder != null)
                ret = new LineIterator(this, url.openStream(), returnNullUponEof, charsetDecoder);
            else if (charset != null)
                ret = new LineIterator(this, url.openStream(), returnNullUponEof, charset);
            else if (charsetname != null)
                ret = new LineIterator(this, url.openStream(), returnNullUponEof, Charset.forName(charsetname));
            else
                ret = new LineIterator(this, url.openStream(), returnNullUponEof, (Charset)null);
            synchronized (openedIterators) {
                openedIterators.add(ret);
            }
            return ret;
        } catch (IOException e) {
            throw new IllegalStateException(e);
        }
    }
    
    public void close() {
        final LineIterator[] lineIterators;

        synchronized (openedIterators) {
            lineIterators = openedIterators.toArray(
                                        new LineIterator[
                                             openedIterators.size()]);
            for (Iterator<LineIterator> itr=openedIterators.iterator(); itr.hasNext();) {
                itr.next();
                itr.remove();
            }
        }
        for (LineIterator li : lineIterators)
            li.closeInPrivate();
    }

    public int numberOfopenedIterators() {
        return openedIterators.size();
    }

    void removeLineIterator(LineIterator li) {
        synchronized (openedIterators) {
            openedIterators.remove(li);
        }
    }

    public boolean isReturnNullUponEof() {
        return returnNullUponEof;
    }

    public void setReturnNullUponEof(boolean returnNullUponEof) {
        this.returnNullUponEof = returnNullUponEof;
    }

    public TextIterable withReturnNullUponEof(boolean returnNullUponEof) {
        setReturnNullUponEof(returnNullUponEof);
        return this;
    }

    public synchronized Charset getCharset() {
        return charset;
    }

    public synchronized void setCharset(Charset charset) {
        this.charset = charset;
        this.charsetname = null;
        this.charsetDecoder = null;
    }
    
    public TextIterable withCharset(Charset charset) {
        setCharset(charset);
        return this;
    }

    public synchronized CharsetDecoder getCharsetDecoder() {
        return charsetDecoder;
    }

    public synchronized void setCharsetDecoder(CharsetDecoder charsetDecoder) {
        this.charsetDecoder = charsetDecoder;
        this.charsetname = null;
        this.charset = null;
    }
    
    public TextIterable withCharsetDecoder(CharsetDecoder charsetDecoder) {
        setCharsetDecoder(charsetDecoder);
        return this;
    }

    public synchronized String getCharsetname() {
        return charsetname;
    }

    public synchronized void setCharsetname(String charsetname) {
        this.charsetname = charsetname;
        this.charset = null;
        this.charsetDecoder = null;
    }
    
    public TextIterable withCharsetname(String charsetname) {
        setCharsetname(charsetname);
        return this;
    }
}

/**
 * @author Hanson Char
 */
@NotThreadSafe
class LineIterator implements Iterator<String>, Closeable {
    private boolean hasNextExecuted;
    private String line;
    private LineNumberReader lnr;
    private final TextIterable textIterable;
    private final boolean returnNullUponEof;

    LineIterator(TextIterable textIterable, InputStream is, boolean returnNullUponEof, Charset charset) {
        this.textIterable = textIterable;
        this.returnNullUponEof = returnNullUponEof;
        InputStreamReader isr = null;

        try {
            isr = charset == null 
                ? new InputStreamReader(is) 
                : new InputStreamReader(is, charset)
                ;
            lnr = new LineNumberReader(isr);
        } catch (Exception ex) {
            try {
                if (lnr != null)
                    lnr.close();
                else if (isr != null)
                    isr.close();
                else if (is != null)
                    is.close();
            } catch (Throwable ignore) {
            }
        }
    }
    
    LineIterator(TextIterable textIterable, InputStream is, boolean returnNullUponEof, CharsetDecoder decoder) {
        this.textIterable = textIterable;
        this.returnNullUponEof = returnNullUponEof;
        InputStreamReader isr = null;

        try {
            isr = new InputStreamReader(is, decoder);
            lnr = new LineNumberReader(isr);
        } catch (Exception ex) {
            try {
                if (lnr != null)
                    lnr.close();
                else if (isr != null)
                    isr.close();
                else if (is != null)
                    is.close();
            } catch (Throwable ignore) {
            }
        }
    }

    public boolean hasNext() {
        if (hasNextExecuted)
            return line != null;
        try {
            hasNextExecuted = true;

            if (lnr != null) {
                line = lnr.readLine();

                if (line == null)
                    close();
            }
            return line != null;
        } catch (IOException e) {
            throw new IllegalStateException(e);
        }
    }

    public String next() {
        if (hasNextExecuted) {
            hasNextExecuted = false;
            return line == null 
                 ? eof() 
                 : line
                 ;
        }
        return hasNext() 
             ? next() 
             : eof()
             ;
    }
    
    private String eof() {
        if (returnNullUponEof)
            return null;
        throw new NoSuchElementException();
    }

    public void close() {
        if (lnr != null) {
            textIterable.removeLineIterator(this);
            closeInPrivate();
        }
    }
    
    public int getLineNumber() {
        return lnr == null  
             ? -1 
             : lnr.getLineNumber()
             ;
    }
    
    void closeInPrivate() {
        if (lnr != null) {
            try {
                lnr.close();
            } catch (IOException ignore) {
            }
            line = null;
            lnr = null;
        }
    }

    public void remove() {
        throw new UnsupportedOperationException("remove not supported");
    }

    @Override
    public void finalize() {
        try {
            super.finalize();
        } catch (Throwable ex) {
        }
        close();
    }
}

Nice command line parser

2007-10-01T10:59:00.000-07:00

Other interesting ones here.

Epigrams In Programming

2007-09-26T21:42:00.000-07:00

by Alan J. Perlis

Perfection

2007-09-26T19:23:00.000-07:00

Perfection is achieved, not when there is nothing more to add, but when there is nothing left to take away. - Antoine de Saint-Exupery (1900-1944)

Parametric Initialization On Demand Holder Idiom

2007-09-01T20:11:00.000-07:00

Similar to the Initialization On Demand Holder (IODH) Idiom , it occurs to me that sometimes not only is there a need to access a singleton instance, which is lazily initialized if constructed for the first time, but also to initialize with parameters, which are ignored if the singleton instance has already been constructed. Also, if there is concurrent access to the singleton instance for the very first time, it doesn't matter from which thread the parameters are passed, as long as the parameters are specified by one of these threads (ie the parameters don't come "out-of-thin-air".)

The API would be something like:

   SomethingMore singleton = SomethingMore.getInstance(params);

Now this begs the question: similar to the original IODH Idiom, can the implementation of such Parametric IODH Idiom be thread-safe without explicit synchronization ?

See below for 2 proposed implementations. One requires the parameter to be immutable, and the other requires the parameter to be only thread-safe.

/**
 * Parametric Initialization On Demand Holder Idiom 1
 * - requires parameter to be immutable and thread-safe
 * even in the face of unsafe publication.
 * (Usually such immutability can be achieved via final fields.)
 *  
 * @author Hanson Char
 */
public class SomethingMore 
{
    /** 
     * An immutable value,
     * such as a {@link String} instance. 
     */
    private final Object value;
    private SomethingMore(Object value) { this.value = value; }
    public Object getValue() { return value; }

    /** 
     * A temporary buffer to hold the immutable value used to initialize 
     * the singleton instance of SomethingMore.
     * Note it is an unsafe publication, but that's ok as 
     * the immutable value passed in is required to be thread-safe even
     * in the face of unsafe publication.
     */
    private static Object valueHolder;
    private static class LazySomethingMoreHolder {
        public static SomethingMore something = new SomethingMore(valueHolder);
    }
    
    /**
     * Returns the singleton instance of {@link SomethingMore}.
     * 
     * @param value an immutable value used to initialize the 
     * singleton instance of SomethingMore, if the instance is 
     * constructed for the first time;  
     * ignored otherwise.  
     * Note immutability (ie lack of mutator methods) 
     * itself is not sufficient.
     * The immutable value must be thread-safe even
     * in the face of unsafe publication.
     * (Usually such immutability can be achieved via final fields.)
     * Caller of this method is responsible for passing in a
     * value which is immutable and thread-safe even in the
     * face of unsafe publication,
     * such as a {@link String} instance. 
     */
    public static SomethingMore getInstance(Object value) {
        SomethingMore.valueHolder = value;
        return LazySomethingMoreHolder.something;
    }
}

/**
 * Parametric Initialization On Demand Holder Idiom 2
 * - requires parameter to be thread-safe.
 * 
 * @author Hanson Char
 */
public class SomethingMoreSafe 
{
    /** A thread-safe value, such as a {@link String} instance. */
    private final Object value;
    private SomethingMoreSafe(Object value) { this.value = value; }
    public Object getValue() { return value; }
    
    private static final ThreadLocal tlocal = new ThreadLocal();
    private static class LazySomethingMoreSafeHolder {
        public static SomethingMoreSafe something = new SomethingMoreSafe(tlocal.get());
    }

    /**
     * Returns the singleton instance of {@link SomethingMoreSafe}.
     * 
     * @param value a thread-safe value used to initialize the 
     * singleton instance of SomethingMoreSafe, if the instance is 
     * constructed for the first time;  
     * ignored otherwise.
     * Caller of this method is responsible for passing in a
     * value which is thread-safe, 
     * such as a {@link String} instance. 
     */
    public static SomethingMoreSafe getInstance(Object value) {
        tlocal.set(value);
        try {
            return LazySomethingMoreSafeHolder.something;
        } finally {
            tlocal.remove();
        }
    }
}

Special thanks to Joe Bowbeer, David Holmes, Dhanji R. Prasanna and Jeremy Manson for their help in the JSR-166 concurrency forum. More discussion can be found here.

SQL*Plus column formatting

2007-07-25T01:03:00.000-07:00

column column_name format a50 word_wrapped

pgrep, ps, etc.

2007-06-26T14:53:00.000-07:00

pgrep -lf jdk
# Starting time or date of a process
ps -o start_time <pid>
# Elapsed time of a process
ps -o "%t" <pid>

OpenSsl/Keytool Cheat Sheet

2007-06-18T15:31:00.000-07:00

# Convert an RSA public key from PEM to DER
openssl rsa -inform PEM -in test-public-key.pem -outform DER -out test-public-key.der -pubin

# Convert a X509 certificate from PEM to DER
openssl x509 -in test-x509-cert.pem -inform PEM -out test-x509-cert.der -outform DER

# Print a X509 certificate
openssl x509 -text -in test-public-cert.pem

# Import a certificate to JKS
jdk1.6.0/bin/keytool -storepass changeit -keystore truststore.jks -importcert -alias mycert -file mycert.der -trustcacerts

# http://www.agentbob.info/agentbob/79.html
# Read the javadoc in ImportKey.java for details of converting it to a keystore

# Convert a private key from PEM to DER
openssl pkcs8 -topk8 -nocrypt -in private-key.pem -out private-key.der -outform der

# Convert cert from PEM to DER
openssl x509 -in cert.pem -out cert.der -outform der

# Convert private key and cert to keystore
java -cp . ImportKey private-key.der cert.der

# Change the storepass
keytool -keystore keystore.jks -storepass importkey -storepasswd -new changeit

# Change the alias
keytool -keystore keystore.jks -storepass changeit -changealias -alias importkey -keypass importkey -destalias myhost

# Change the keypass
keytool -keystore keystore.jks -storepass changeit -alias myhost -keypasswd -keypass importkey -new changeit

# List the keystore
keytool -keystore keystore.jks -storepass changeit -list -v

# How to generate a self-signed key pairs using keytool ?
keytool -genkeypair -dname "cn=myname,ou=myunit,o=myorg,c=AU" -alias myalias -keypass changeit -keystore ./my-keystore.jks
-keypass changeit -storepass changeit -validity 9999 -v

### Extracting the certificate (public key) ###

# Export the X509 certificate
keytool -export -alias myalias -keystore my-keystore.jks -storepass changeit -file mycert.der

# Display it
openssl x509 -noout -text -in mycert.der -inform der

# Convert to PEM
openssl x509 -out mycert.pem -outform pem -in mycert.der -inform der

### Extracting the private key ###

#Download, compile & run ExportPriv.

# Export private key into pcks8 format
javac ExportPriv.java
java ExportPriv my-keystore.jks myalias changeit > my-key.pkcs8

# Combine public and private key into pkcs12 format
openssl pkcs12 -export -out my-key.p12 -inkey my-key.pkcs8 -in my-cert.pem

# Convert pkc12 to PEM so it can be displayed
openssl pkcs12 -in pkcs-12-certificate(-and-key-file) -out pem-certificate(-and-key-file)

# Find out the MD5 of an X509 cert
openssl x509 -fingerprint -md5 -in cert.pem

Hacking JGroups Util.java

2007-06-13T09:38:00.000-07:00

Recently I encountered a resource loading problem in JGroups from within a Tomcat webapp when trying to construct a JChannel when so triggered from a JMX MBean via the jconsole:

    channel = new JChannel("tcp.xml");

Basically if the JChannel is constructed from the webapp, it can locate the "tcp.xml" as a resource and works just fine. However if the JChannel is constructed from an MBean call stack via the jconsole, it failed to locate the resource due to the fact that the context classloading of the executing thread is different. So I hacked org.jgroups.util.Util to make it work in both cases. I made the hack on jgroups 2.2.9.4 , but I tried the latest 2.5.0CR1 which also seemed to exhibit the same problem/behavior. More details below.

My question is: should the hack I made be incorporated into the official version ? Guess what ? The next day, Bela Ban, the JGroups Lead / JBoss Clustering team, unconditionally accepted this change into the official JGroups 2.4 branch and the CVS head. Yeah!

public class Util {
...
    public static InputStream getResourceAsStream(String name, Class clazz) {
        ClassLoader loader;
                                                                                                                   
        try {
            loader=Thread.currentThread().getContextClassLoader();
            if(loader != null) {
// hack: returns only if resource is found
//                return loader.getResourceAsStream(name);
              InputStream is = loader.getResourceAsStream(name);
                                                                                                                   
              if (is != null)
                  return is;
            }
        }
        catch(Throwable t) {
        }
                                                                                                                   
        if(clazz != null) {
            try {
                loader=clazz.getClassLoader();
                if(loader != null) {
// hack: returns only if resource is found
//                    return loader.getResourceAsStream(name);
                    InputStream is = loader.getResourceAsStream(name);
                                                                                                                   
                    if (is != null)
                        return is;
                }
            }
            catch(Throwable t) {
            }
        }
        try {
            loader= ClassLoader.getSystemClassLoader();
            if(loader != null) {
                return loader.getResourceAsStream(name);
            }
        }
        catch(Throwable t) {
        }
                                                                                                                   
        return null;
    }
    ...
}

Two-Phase-Commit Activation Protocol

2007-06-11T11:57:00.000-07:00

Continuing on the previous post on "Guaranteed db activation in all JVM's before table unlock in HA-JDBC ?"...

Actually, I think just doing a synchronous notification is not sufficient. What is required is an atomic/transactional activation of the specified db in all nodes - either all commit or rollback.

This naturally brings up the thought of using 2PC (Two Phase Commit) protocol. Here is a proposal. Say we have a synchronizing node S(ync), and two other members F(oo) and B(ar):

2PC Activation Protocol

(Assuming tables in the active db are locked, and synchronization has just completed and S is now trying to atomically/transactionally activate the inactive db in all JVM's before unlocking the tables.)

From S's perspective:

S synchronously sends an "ActivationReady" message to F and B. If all such sending failed, simply aborts this 2PC activation procedure (ie inactive db stays inactive). If some but not all such sending failed, S sends an "ActivationAbort" message to others that it has previously successfully sent the ActivationReady message, and then simply aborts this 2PC activation procedure (ie inactive db stays inactive). Else proceed to (2).

S synchronously sends an "ActivationCommit" message to F and B and then always activates the inactive db locally (regardless of whether any or all of the sending was successful.)

(Note that S always unlocks the tables in the active db afterwards, regardless of whether the 2PC activation has been aborted or not. However, if S detects any possible failure during commit or rollback, it must wait for the other nodes to timeout before unlocking the tables.)

From F or B's perspective:

If it receives an ActivationReady message followed by an ActivationAbort message, do nothing.
If it receives an ActivationReady message followed by an ActivationCommit message, activate the specified db locally.
In the extreme case that after an ActivationReady message is received but failed to received a subsequent ActivationCommit or ActivationAbort message received within, say, 5 secs, deactivate all db's locally. (This is to prevent any possible compromise to data consistency caused by not knowing whether the db in concern has been activated in other JVM's or not.)

Too complex ? Well, Paul Ferraro, the creator of HA-JDBC, seems to think so:

"I think you are over-engineering this a bit...

Database activation does not benefit from two-phase commit (2pc). The purpose of 2pc is to provide distributed consistency to a fallible operation. The operation in this case, activating a database, is not fallible - it is merely the insertion of a element into a collection, i.e. Balancer.add(...).

If you are curious about an instance where 2pc is warranted, check out the DistributableLockManager in HA-JDBC 2.0, which utilizes the JGroups TwoPhaseVotingAdapter."

But then what should we do in the case when the synchronous activation notification (presuming it has been implemented) to other JVM's resulted in partial success ? In other words, some get notified (and therefore have the inactive db activated) and some not ?

Wouldn't that result in data inconsistency (since some JVM which are succesfully notified would start to write to all the active db's in the cluster, whereas some JVM which are not succesfully notified would continue to write to only some but not all active db's in the cluster) ?

Some interesting comments from Bela Ban, the JGroups Lead/ JBoss Clustering team:

"Yes, if you want atomicity guarantees, you either have to use a uniform protocol (a message is only delivered if all receivers acked it, otherwise it will not get delivered at anyone) or 2PC as you outlined below. Uniform delivery is not part of JGroups, although it is on the todo list (http://jira.jboss.com/jira/browse/JGRP-138). I once had an impl, but it was too slow so I trashed it.

2PC is probably the way to go. I've done this in JBossCache: if you look at the code that is run when you have (a) a transaction and (b)synchronous method calls, then I pretty much run the classic PREPARE --> success? COMMIT : ROLLBACK pattern.

What you describe is such an implementation, but I guess that's something to be discussed in the scope of HA-JDBC. JGroups just provides the reliable transport with retransmission (lossless transport) and ordering."

More tips from Bela Ban:

"The RequestCorrelator is used by both MessageDispatcher and RpcDispatcher, it is at a lower abstraction level. I would suggest RpcDispatcher (for sync or async cluster-wide method calls) or MessageDispatcher (for sync or async cluster-wide message sending). Note that RpcDispatcher uses MessageDispatcher which uses RequestCorrelator.

In both RpcDispatcher.callRemoteMethods() and MessageDispatcher.castMessage() you get back an RspList which contains the result per member:
The result (null if method was void)
Whether the member was suspected, e.g. if you invoke foo() on {A,B,C,D}, and C crashes while you wait for acks, then C will be flagged as suspected in the response list
Whether we received an ack (if the call is bounded by a timeout)
Paul, I'd suggest you use an {Rpc/Message}Dispatcher directly rather than NotificationBus. NB was designed to be a simple asynchronous notification mechanism."

What do you think ?

Guaranteed db activation in all JVM's before table unlock in HA-JDBC ?

2007-06-11T11:41:00.000-07:00

Say we have multiple JVM members accessing a HA-JDBC database cluster which involves a table locking synchronization strategy. Currently after a (inactive) db is synchronized, the code in ha-jdbc 1.1.11, method

    private void LocalDatabaseCluster#activate(Database, List, SynchronizationStrategy) 
        throws java.sql.SQLException

would activate the inactive db (both in the local JVM and other JVM's via jgroup notification) before executing a rollback which would unlock the tables in the active db.

My question is that when the JGroups notification of the activation to other JVM's is returned, does it mean
a) all other JVM's have successfully received the notification, or
b) some (or none or all) of the other JVM's have successfully received the notification, or
c) whether this only means the notification has been sent to other JVM's which may or may not receive the notification ?

If it is (b) or (c), is there any way to achieve/guarantee (a) ? The reason is that, for data consistency purposes, I am trying to provide the guarantee that all JVM's must have the (inactive) database activated before the table locks are released, or otherwise I'd rather to have the activation fail (ie inactive db stays inactive).

Unfortunately, at least for now, it turns out the answer is (c), as quoted from Paul Ferraro, the creator of HA-JDBC, at the ha-jdbc-user forum:

"Excellent question...

The distributed cluster state is implemented using a JGroups NotificationBus. Database activations and deactivations trigger appropriate notification events to listening members. I had always thought that NotificationBus operated synchronously, i.e. that NotificationBus.sendNotification(...) would not return until all nodes complete their NotificationBus.Consumer.handleNotificaiton(...) listener methods. I just completed a test against both JGroups 2.2.9.4 and 2.5-beta-2 - unfortunately, this is not the case. A synchronous messaging bus requires the use of a RequestCollator, and is not used by NotificationBus. I don't remember at what point I made this assumption - but I was wrong.

So the answer to your question - although my intention was (a), currently, it is (c). This needs to change. I need to re-implement DistributableDatabaseCluster (DistributableStateManager, in 2.0) to use a MessageDispatcher, DistributedQueue, or some other construct that utilizes a RequestCollator...."

Java Debug Interface

2007-06-06T23:33:00.001-07:00

Just realize JDI is such a powerful interface with implementation available in the jdk's tools.jar. Imagine you can write your own code to launch other VM's or attach to a running VM, inspect the executing threads, print out what methods the target VM is executing, etc. Essentially, with JDI, you can programmatically do whatever a debugger can with ease!

See here for some interesting sample code built on JDI.
See here for an interesting article by Gilad Bracha and David Ungar about Meta-Programming facilities somehow related to JDI.

Heap Dump in Java 6

2007-05-21T00:39:00.001-07:00

Just experimenting with the use of jmap and jhat in Java 6. The Object Query Language in jhat that allows the use of javascript closure seems pretty powerful, though also pretty slow.

Practices
# Find out the jvm pid
jps
# Trigger heap dump
jmap -dump:format=b,file=/tmp/java_app-heap.bin <pid>

# Analyzing the heap dump
jhat -J-Xmx326m /tmp/java_app-heap.bin

http://localhost:7000/showInstanceCounts/
http://localhost:7000/oql

#Find out the total size of MyClass instances
select sum(map(heap.objects('foo.bar.MyClass'), 'sizeof(it)'))

# List out all class names matching "foo.bar"
select filter(heap.classes(), "/foo.bar/(it.name)")

# Ditto
select map(filter(heap.classes(), "/foo.bar/(it.name)"),
    function(c) {
      return "<br>"+ c.name;
    })

# List classes that match "foo.bar" with instances > 0
select map(filter(heap.classes(), "/foo.bar/(it.name)"),
    function(c) {
      var len = length(heap.objects(c));
      if (len == 0) return "";
      return "<br>" + c.name + ", instances:" + len;
    })

# List classes that match "foo.bar" with instance sizes > 0
select map(filter(heap.classes(), "/foo.bar/(it.name)"),
    function(c) {
      var totalsize = sum(map(heap.objects(c), 
          function(o) {
              return sizeof(o);
          }
      ));
      if (totalsize == 0) return "";
      return "<br>" + c.name + ", size:" + totalsize;
    })

More info
How to find java memory leaks
Troubleshooting Guide for Java SE 6 with HotSpot VM
HeapDumpOnOutOfMemoryError option in 5.0u7 and 1.4.2_12
Permanent generation analysis with OQL
What's in my Java heap?
Querying Java heap with OQL
JavaOne 2007 BOF on Memory Leaks
BOF9982: The java.lang.OutOfMemoryError: PermGen Space error demystified

CVS Access to JSR166

2007-05-02T22:18:00.000-07:00

Here is the anonymous CVS access to JSR166:

:pserver:anonymous@gee.cs.oswego.edu:/export/home/jsr166/jsr166

with project also named jsr166 (as in "cvs co jsr166"). Wouldn't it be nice if this information is included at the jsr166 web page,

http://g.oswego.edu/dl/concurrency-interest/

in addition to the browsable CVS links ?

Thanks to Doug Lea.

Oracle Table Differential

2007-04-05T15:13:00.000-07:00

Say you have two Oracle databases, db1 and db2, that contain the same schema. Every table T in db1 should contain the same records as that in db2 . Also, every table only allows read and insert operations (ie no updates or deletes), and every record has a creation date.

Now for some reasons db2 went down. Meanwhile there are continuously new records being inserted into T in db1. When db2 is back up, how do we figure out the set of records that need to be inserted back to T in db2 ?

Note you cannot assume db1 and db2 has the same clock time, but you can make the hypothetical assumption that the creation time of all the records in a specific table are in total chronological order. (In reality multiple records can be created with the same creation date/time, but let's get to that later.)

Here is one way of doing it. Let's call it the Differential Procedure:

db2Count = select count(*) from db2.T;
db1Count = select count(*) from db1.T;
diff = db1Count - db2Count;

select * from (
   select * from (
       select * from (select * from db1.T order by creation_date) 
       where rownum <= db1Count 
       order by creation_date desc))
where rownum <= diff 
order by creation_date;

Can you think of a better/faster one ?

Updated on 23Oct - Actually the above SQL can be simpler:

select * from (
   select * from (select * from db1.T order by creation_date) 
   where rownum <= db1Count 
   order by creation_date desc)
where rownum <= diff 
order by creation_date;

Now back to reality. How do we handle those rows that got inserted into db1 as at the time when db2 went down ? They all have the same creation date in db1 but some of them may exist in db2 and some may not! Oh well, how about this:

Figure out all those rows that were created as at the time when db2 went down:

select * from db1.T where creation_date = (
 select max(creation_date) from (
   select creation_date from db1.T order by creation_date) where rownum <= db2Count)

For each of these rows, insert it to db2 only if it doesn't already exist;

Now apply the Differential Procedure to db1 and db2. This time, however, we got a set of records that must all exist in db1 but not in db2. Yum! Simply insert them all to db2.

Finally, if you are like me thinking about using HA-JDBC to cluster db1 and db2, how can we activate/synchronize db1 and db2 on the fly with minimal impact on the clients ? In other words, can synchronization be done correctly without the tables being locked ?

For example, take the hypothetical case above and say table T in db2 had 10 rows when it was deactivated. When db2 becomes ready for activation, db1 may have already got 100 rows. So to sync up, the extra 90 needs to be inserted in db2. Yet once that's done, db1 may have gone up to 110. The problem is db1 may just keep growing!

Is there a clever way to get around this dynamic changing synchronization problem yet without resorting to the use of table locking ? I can't really think of one :(

But how about this:

Repeatedly apply a non-locking synchronization procedure, such as the Differential Procedure, until the delta of table T between the active and inactive db falls below a certain threshold;
Lock table T of all active databases
Apply the synchronization procedure to T for the last time
Active the inactive db
Release the table locks. (Just like what LocalDatabaseCluster#activate does.)

There will still be impact on the clients, but the lock duration is minimized.

What do you think ?

ssh tunnelling fun

2007-03-24T23:44:00.000-07:00

To access a service that only accepts requests at localhost:9090 at myserver, I can ssh tunnel from mypc to myserver:

ssh -L 9090:localhost:9090 -N myserver

Now, access to localhost:9090 at mypc would be forwarded to localhost:9090 at myserver.

More info on ssh tunnelling here.

Another example:

Say host-A can access host-B whereas mypc can only ssh to host-A but cannot access host-B. To access host-B from mypc:

ssh -N -p 22 -c 3des userid@host-A -L 9090/host-B/9090

Now, access to localhost:9090 at mypc would be forwarded to host-B:9090

jconsole ssh tunnelling

Now say one can normally jconsole via a service url like:

jconsole service:jmx:rmi://myhost:6160/jndi/rmi://myhost:6110/server

But if ports 6160 and 6110 are restricted, it would render jconsole almost useless. Here is how ssh tunnelling may come to rescue (if you are allowed to ssh to the host):

ssh -L 6160:localhost:6160 -N myhost
ssh -L 6110:localhost:6110 -N myhost
jconsole service:jmx:rmi://localhost:6160/jndi/rmi://localhost:6110/server

Oh, don't forget to start up the JVM with

    -Djava.rmi.server.hostname=localhost

Sequoia 2.10.6

2007-03-19T14:31:00.000-07:00

Ever tried the Sequoia 2.10.6's raidb1 demo ? Here is something interesting: once the demo's local jmx server has been started by the Sequoia controller, you can peek into it via jconsole:

jconsole service:jmx:rmi://localhost/jndi/rmi://localhost:1090/jrmp

and also:

jconsole service:jmx:rmi://localhost/jndi/rmi://localhost:1091/jrmp

if you are running the raidb1 distributed demo.

Photos@Belgium

2006-12-20T22:26:00.000-08:00

JavaPolis 2006

2006-12-14T13:49:00.000-08:00

This year the theme of JavaPolis is something like there is a better place to meet your idols. Oh well, I certainly did meet up with some of my idols. I could even ask a few questions or have a conversation with some of them face-to-face!

Brian Goetz - The Java Concurrency hero. I asked him on the expressiveness of the concurrency annotations in JCIP, and whether the JDK should be annotated on thread safety. He pointed me to JSR305.

Mark Reinhold - Gave an interesting talk on integrating XML into the Java Language. Apparently he has quite a few ideas about how this can be done in Java 7. However, from some informal conversations, it seems Neal was not so sure about having such integration at the language level, as API such as JDOM should suffice to tackle the pain points.

Neal Grafter - I asked Neal whether the Tree/Event hybrid XML streaming API presented by Mark Reinhold would degenerate to the behavior of a DOM API if a programmer carelessly matches the root document element. Neal kindly agreed. His presentation on Closure was as usual deep and thought provoking.

Dr. Heinz - I finally got the chance to talk to Dr. Heinz, the Java Specialist, whom I have been exchanging emails with in the past. An interesting and friendly guy with some serious insights into Java. We had some Belgium fries together. Yum.

Bill Venners - Talked about how to deal with software exceptions. I asked him if a condition is considered something that should never occur, whether throwing an AssertionError would be more preferable to a RuntimeException. He was not sure, and was honest about his uncertainty.

Marc Fleury - Very entertaining guy. For some reasons he reminded me of Rocky Balbor. He gave an interesting presentation of different open source models, emphasized the importance of a brand name, and his personal advocate of the use of GPL/dual license for startup based on professional open-source.

Erich Gamma - Gave an interesting presentation about the Eclipse development process, how it moved from a totally closed development to completely open-source, and how transparency can help foster the health of a development process. Something to learn about.

And there is free Belgium beer. A remarkable conference.

Beanlib at JavaPolis 2006

2006-12-13T16:30:00.000-08:00

I did a little quickie presentation on the Beanlib opensource project at JavaPolis 2006. I'll try to post a link to the presentation here later on.

Bamboo - Opensource DHT

2006-10-27T13:54:00.000-07:00

Looks interesting.

Jaxb2 jakarta-commons-lang Plugin

2006-09-18T19:15:00.000-07:00

Just contributed to the reference implementation of Jaxb2 a very simple plugin named jakarta-commons-lang plugin that would add toString(), hashCode() and equals() method to the xjc generated Pojo's using the Jakarta Commons' ToStringBuilder, HashCodeBuilder and EqualsBuilder.

More Jaxb2 plugins can be found here. Enjoy!

WriteThottleFilter in Mina

2006-09-13T06:54:00.000-07:00

I was wondering if there is an easy way in Mina 0.8.2 to control the rate of physical network write operations to not exceed a specific number of messages.

For example, can I say for a particular socket connection session Mina should write physically no more than 3 messages per second ? Message here means the argument that get passed to ProtocolSession.write(Object).

AFAIK the ProtocolSession.write() is a logical asyn operation, and Mina therefore reserves the right to buffer it underneath. If so, controlling the rate of session.write() cannot be relied upon to control the rate of physical messages that get sent down the wire.

So I posted the question to the Mina developer forum, and here is the reply by Trustin Lee:

"You are correct, but you can make session.write() a blocking operation by inserting a filter which limits the write rate. Please implement your filter's filterWrite() to throttle the write request.

BTW this idea is nice. Could you please create a JIRA issue for us so we can resolve it someday and you can switch over to our version of write rate limiting filter? Otherwise you could contribute! ;)"

It turns out the implementation can be quite simple:

...
    SocketConnector socketConnector = new SocketConnector();
    socketConnector.getFilterChain()
                   .addLast("writeThrottleFilter",
                            new WriteThrottleFilter(333));
... 

public class WriteThrottleFilter extends IoFilterAdapter 
{
    /** Delay in milli-seconds between writes. */
    private final long delayMillis;
    
    public WriteThrottleFilter(long delayMillis) {
        this.delayMillis = delayMillis;
    }
    
    @Override
    public synchronized void filterWrite(NextFilter nextFilter,
            IoSession session, ByteBuffer buf, Object marker) 
            throws InterruptedException
    {
        nextFilter.filterWrite( session, buf, marker );
        Thread.sleep(delayMillis);
    }
}

Thanks to Trustin Lee, this class will be included in the Mina 1.0 release! More details here.

ConcurrentLinkedBlockingQueue

2006-09-10T23:52:00.000-07:00

I've been wondering why there is ConcurrentLinkedQueue in Java 5+, but not something like a ConcurrentLinkedBlockingQueue, which would allow the client to block on an empty queue via a "take" method, or block on an empty queue for a limited time via a "poll" method.

I attempted to construct one by combining the use of a Semaphore with a ConcurrentLinkedQueue, but apparently it is less scalable than the LinkedBlockingQueue.

As pointed out by Doug Lea:

"This IS a good thought....This way works, but reduces concurrency by using a single semaphore, so is a bit less scalable than current inkedBlockingQueue. However, there is a path to much better scalability by using the "dual-queue" approach similar to what we did for Java 6 SynchronousQueue. My initial intent was to find a way to internally use such techniques to replace the unbounded case of LinkedBlockingQueue. But this turns out not to work out too well because of all the little compatibility problems encountered -- for example, maintaining the same Serialization form. So it is more likely that we'll put out a separate ConcurrentLinkedBlockingQueue that will be preferable to LinkedBlockingQueue unless you need capacity constraints.
Stay tuned for it..."

In stead of waiting, I decided to try a different path and proceeded to make a second attempt. This time the constructs I use included:
* an extra ConcurrentLinkedQueue for the parking threads
* volatile for marking if a thread is potentially parked or not
* LockSupport.[un]park[Nano](...)

One feature of such ConcurrentLinkedBlockingQueue is that it is unbounded, whereas even LinkedBlockingQueue has a max capacity of Integer.MAX_VALUE.

The source can be found here, whereas some performance tests can be found here. Some initial load testing looks promising, although I only tested it with a single procesor machine, and it appears the test result is very sensitive to the GC and heap memory configuration for the JVM. What I really need is some loading testing on a multi-processor environment where true concurrency is possible.
Update on 11Sep2006: Running the test harness on an AMD Opteron dual processor using JDK-1.5.0_05, the CLBQ is on average 72% faster than the LBQ. See here for more details.

Update on 19Sep2006:

More experiments seem to indicate that the CLBQ consistently outperform the LBQ for the mix of N-producer 1-consumer. When there are N-producer and M-consumer, however, it seems the reverse is true. Not exactly sure why. Probably CLBQ incurs a slightly higher overhead as compared to LBQ in the M-consumer situation.

Update on 4Apr2007:

If anyone has a powerful box (with multi-processors) and is interested in comparing the performance of ConcurrentLinkedBlockingQueue vs LinkedBlockingQueue, please download a jar from:

http://beanlib.sourceforge.net/clbq/070407/q-test.jar

and run it with either jdk5 or jdk6:

java -DnumConsumer=10 -DnumProducer=100 -DwcRatio=20 -server -XX:CompileThreshold=1500 -XX:+UseConcMarkSweepGC -XX:+CMSIncrementalMode -XX:+DisableExplicitGC -jar q-test.jar

Note:

the number of concurrent producers can be specified via the system property "numProducer", which defaults to 10
multiple concurrent consumers is now supported via the system property "numConsumer", which defaults to 1
the W/C ratio (ie wait-time to compute-time) can be specified via the system property "wcRatio", which defaults to zero

For instance, the above example means to run the test with 100 producers, 10 consumers, and a W/C ratio of 20. One interesting observation is that Windows XP Professional is really slow on LBQ even when there is only 1 producer and 1 consumer! Feel free to tinkle with the parameters, and let me know what you find.

The source jar of the test harness can be downloaded from:

http://beanlib.sourceforge.net/clbq/070407/q-test-sources.jar

jsr166x in Java 5

2006-08-26T23:24:00.000-07:00

Update on 27Aug06: It turns out including ConcurrentLinkedDeque is not such a good idea, and therefore removed in Beanlib 3.2.5. As pointed out by Doug Lea, this class will be replaced with something much better in Java 7.

Ever thought of taking advantage of those concurrent classes such as ConcurrentSkipListMap in Java 6 beta, but still being trapped in a Java 5 environment ? Well the entire jsr166x package has now been ported to beanlib 3.2.4, with sources augmented with the latest from Java 6 beta 2. This included even the ConcurrentLinkedDeque which is currently missing in Java 6.

Enable Remote Debugging + JMX, etc. from Ant

2006-08-03T13:00:00.000-07:00

Sample configuration:

  <target name="mytarget">
      <java classname="my.packagename.MyMainClassName"
        fork="yes"
      >
          <!-- For remote debugging -->
          <sysproperty key="DEBUG" value="true" />
          <jvmarg value="-Xrunjdwp:transport=dt_socket,server=y,suspend=y,address=8000" /> 
          <!-- For enabling JMX monitoring -->
          <sysproperty key="com.sun.management.jmxremote.port" value="8001" />
          <sysproperty key="com.sun.management.jmxremote.authenticate" value="false" />
          <sysproperty key="com.sun.management.jmxremote.ssl" value="false" />
          <!-- GC tuning. -->
          <jvmarg value="-verbose:gc" />
          <jvmarg value="-server" /> 
          <jvmarg value="-XX:+UseParallelGC" />
          <jvmarg value="-XX:GCTimeRatio=19" />
          <!-- Heap tuning. -->
          <jvmarg value="-Xms512m" />
          <jvmarg value="-Xmx512m" />
          <!-- JNI -->
          <env key="LD_LIBRARY_PATH" value="/my/so/directory" />
          <!-- Arguments, if any -->
          <arg value="my.argument"/>
      </java>
      <classpath>
          <pathelement path="${sample.dir}"/>
          <path refid="sample.classpath"/>
      </classpath>
  </target>

Nice resource for setting up Remote Debugging for Java.
Nice articles on tuning GC can be found here, here and here.
Java Ant Task here.

Spring 2.0RC2 is not backward compatible ?

2006-07-06T23:49:00.000-07:00

Update on 8Jul06: Jira issue created here as per Rob Harrop's request.

Once I replaced the 2.0 RC1 spring.jar with the one from 2.0 RC2, I got a NullPointerException.

More details here.

Bug Patch Mina 0.8.2

2006-06-23T08:28:00.000-07:00

Update on 6July06:It turns out the root cause of the problem is related to my Decoder implementation. More details can be found in the JIRA Issue.

The lesson is: Don't use Mina 0.8.2. Go for Mina 0.9.4+ or else risk data corruption.

Update on 4July06: JIRA Issue submitted to the Mina group.

Just found a nasty race condition in using

org.apache.mina.util.Queue

in Mina 0.8.2.

When invoked from IOAdapter$SessionHandlerAdapter.dowrite(), the Queue.isEmpty() is not properly synchronized. This causes data corruption when the network IO is under high load! I was hit by it when I changed my log4j configuration to use pure asyn appenders.

The good news is it's easy to patch. I simply took the gut out of the original Mina's queue implementation, and replaced it with a simple delegation to

java.util.concurrent.ConcurrentLinkedQueue

There are some features lost such as the removal of the List interface (from the Mina's Queue class). But in this case the additional List interface methods are not actually used anywhere else in Mina 0.8.2!

IMHO, Mina should really take advantage of JKD5 features such as non-blocking synchronization , instead of having synchronized statements/blocks all over the places. After all, isn't Mina all about performance ?

(Will try to post the patch back to the Mina group when I got a chance.)

Java Xml Binding

2006-06-16T23:24:00.000-07:00

Thinking of which Java XML binding technology to use for building your
next web services ? JAXB, JiBX, Castor, XmlBean, XStream ?

No more. With Spring Web Services 1.0M1, simply use the same
Marshaller/Unmarshaller interfaces, and the different O/X mapping
implementations can be instantly swapped in and out as you wish! No
brainer. Or at least it seems so.

Latest Jaxb RI 2.0.1 Fluent API Plugin available!

2006-06-11T23:32:00.000-07:00

Due to the specific web service environment in my workplace, I didn't have a chance to touch JAXB RI 2.0.x for quite a while after writing the Fluent API plugin. There is recently an interesting request by Kenny MacLeod about enhancing the plugin to generate fluent setter method with variable arguments for List member fields.

So I spent quite some time to refresh myself as to what exactly I have done, and went tweaking, and tweaking. Finally, the enhanced plugin is now available for download :)

The plugin has also been submitted to maven 2 for distribution. Details here and here.

HalfSync

2006-05-24T16:35:00.000-07:00

It seems we can easily construct something which is simple, thread-safe, and yet competitive to or even faster than j.u.c.a.AtomicInterger!

Consider the code:

// Only the write is synchronized, not the read
public class HalfSync {
    private volatile int count;
   
    public HalfSync(int count) {
        this.count = count;
    }

    public int getCount() {
        return count;
    }

    public synchronized int increment(int delta) {
        return this.count += delta;
    }
}

Interesting comments by David Holmes:

"- reads: same
    both do a LD with whatever memory barrier is needed on the
    platform (which is probably none)
 - writes:
   - uncontended: close call
       AtomicInteger.get has a method call with LD and CAS plus MEMBAR
       Half-sync: CAS for synchronized, LD, ST plus MEMBAR (depends if
                  the runtime elides the redundant MEMBARS for sync+volatile
  - contended:  Half-sync wins by avoiding ctx switches"

j.u.c.atomic.Atomic*.weakCompareAndSet

2006-05-15T08:25:00.000-07:00

I have been wondering what it means when it "fails spuriously" in invoking those weakCompareAndSet methods in Java 5, and what a good use case would be like. It turns out the spurious "failure" simply means the method would do nothing and return false for no apparent reason.

Sample Usage:

public class Foo {
   private volatile int latency = -1;
   private final AtomicInteger maxLatency = new AtomicInteger(-1);

   public void setLatency(int latency)
   {
       this.latency = latency;
       updateMaxLatency(latency);
   }

   private void updateMaxLatency(int latency)
   {
       for (;;) {
           int maxLatencySnapshot = this.maxLatency.get();
         
           if (latency > maxLatencySnapshot)
           {
               if (!this.maxLatency.weakCompareAndSet (maxLatencySnapshot, latency))
                   // race condition or just fail spuriously; so let's retry
                   continue;
           }
           return;
       }
   }
   // ...
}

As commented by Doug Lea:

"Yes, this is fine; it is a good example where either the plain or the weak form would work just as well, so you might as well use the weak form."

Special thanks to David Holmes for his generous and thorough explanation.

Beanlib in maven2

2006-04-20T17:01:00.000-07:00

Thanks to Joe D. Velopar, beanlib is now available in the maven 2 repository.

The original request can be found here, and here is the how-to guide.

Simply add to the pom.xml something like:

  <dependencies>
    ...
    <dependency>
      <groupId>net.sf.beanlib</groupId>
      <artifactId>beanlib</artifactId>
      <version>3.1.1</version>
    </dependency>

    <dependency>
      <groupId>net.sf.beanlib</groupId>
      <artifactId>beanlib-hibernate</artifactId>
      <version>3.1.1</version>
    </dependency>
    ....
  </dependencies>

and your are all set.

Go maven 2!

How to make a software engineer excel ?

2006-04-01T16:13:00.000-08:00

What do you think ? What's the ideal working environment for you as a software engineer ?

Ant built-in properties

2006-03-31T23:16:00.000-08:00

A nice Ant target to print out all the Ant's built-in properties:

<target name="echo" >
    <echo message="os.name:          ${os.name}"          />
    <echo message="basdir:           ${basedir}"          />
    <!-- absolute path of the build file. -->
    <echo message="ant.file:         ${ant.file}"         />
    <!-- path to executing ant's root directory. -->
    <echo message="ant.home:         ${ant.home}"         />
    <echo message="ant.version:         ${ant.version}"   />
    <echo message="ant.project.name: ${ant.project.name}" />
    <echo message="ant.java.version: ${ant.java.version}" />
    <!-- System properties. -->
    <echo message="user.home:         ${user.home}"   />
    <echo message="user.name:         ${user.name}"   />
    <echo message="java.home:         ${java.home}"   />
</target>

Originally found here.

int to bytes

2006-03-19T22:40:00.000-08:00

Here is a fun quiz: What is the best possible way (fastest with least memory) to convert an int to a byte array (of size 4) in big endian in Java ? ie.

public static byte[] toBigEndianBytes(int i) {
// ??
}

It turns out the code below is the best I've found so far:

public static byte[] toBigEndianBytes(int i) {
        return new byte[] {
                (byte)(i >>> 24), 
                (byte)(i >>> 16),
                (byte)(i >>> 8),
                (byte)i
        };
}

Can you think of a better one ?

Update on 21Mar06: Interesting class to look at java.nio.Bits.
Update on 29Mar06: Paul Brown has more details.

Backup.sh

2006-03-11T00:25:00.000-08:00

A simple little script that does the backup for me to a remote machine:

backup.sh
---------
filename=`date +%y%m%d`-`date +%H%M%S`-xyz-projects.tar
echo $filename
tar cf $filename ProjectXyz*
gzip $filename
scp $filename.gz hchar.desktop:/home/hchar/backup/
rm $filename.gz

Windows Subversioned

2006-03-06T12:27:00.000-08:00

Just got SVN server running on Windows. So far so good! I installed both the TortoiseSVN and SVN server using the SVN 1-Click Setup.

It's not obvious how to specify the SVN URL. For example, if the SVN repository is created at C:\svnrepos, the URL will be

file:///C:/svnrepos

I then imported my Java project to it from Eclipse 3.1.2 via the Subclipse plugin. My project is now actually under dual version control by SVN and P4 :)

To remote access it, the SVN URL is simply:

svn://myhostname

Pretty neat.

Update on 12Mar06: just upgraded my svn client to TortoiseSVN 1.3.2 and svn server to SVN 1.3.0 on Windows. I didn't uninstall the previous subversion (both client and server), but I needed to manually disable the SVNService before I could install SVN 1.3.0 without failure.

I've used subversion seriously only for a week. Love it so far. In fact, I've disabled my CVSNT service (although not yet uninstalled). There is even utility to convert existing CVS repository to SVN. So I am not looking back.

Alternative to implementing js as jsp

2006-02-25T09:31:00.000-08:00

Here is an idea that is not as simple as just directly implementing js as jsp, but is probably more performant and maintainable: for each js, we also have a corresponding jsp. For example

Foo.js
Foo.js.jsp

Within Foo.js, it's pure static javascript, but optionally with "hook"'s that the Foo.js.jsp can fill in during runtime on the enclosing page. For example:


Foo.js
------
// Constructor of the Javascript Foo class.
function Foo() {}
...
// Anytime it needs to access a dynamic property that needs 
// to be determined by the Java App, or tag lib, or whatever, 
// use a class level variable.
// For example, a i18n message to be displayed.
alert(Foo.FOO_MESSAGE);
...

Foo.js.jsp
----------
// Foo.js.jsp is included in the main page where the reference to Foo.js is made.
// If there is no dynamic content, Foo.js.jsp just evaluates to do nothing.
Foo.FOO_MESSAGE = "<c:out name="${fooMessage}" />"

This way, all the dynamic content are included as Javascript class level variables in the main page, and the static Foo.js just refers to these variables.

Now all the *.js can be pushed to the Web server front end (eg Apache) - solves the performance issue. All the dynamic content is maintained in one single place: the respective jsp connected via a javascript class - solves the maintenance nightmare issue.

Replace tabs with spaces

2006-02-06T11:03:00.000-08:00

# Replace tabs with 4 spaces in all java files
for i in `find . -name '*.java'`; do sed -ie 's/\t/    /g' $i; done
# Clean up left-over
find . -name '*.javae' -exec rm -f {} \;

Eclipse WTP rocks!

2006-02-01T23:00:00.000-08:00

The Eclipse Web Tools Platform Plugin Project is awesome! Great for developing J2EE applications. I particularly like the visual schema editors for xsd and wsdl files.

Can be easily installed via the Eclipse Update Manager by entering this URL:

http://download.eclipse.org/webtools/updates/

Works like a charm with Eclipse 3.1.2.

How to write a jaxb2 plugin ?

2006-01-30T19:46:00.000-08:00

Subscribe to users@jaxb.dev.java.net;
Check out the JAXB2 RI workspace. See https://jaxb2-sources.dev.java.net/ for the instruction;
Consider looking at the "Developing on top of JAXB RI" section of https://jaxb.dev.java.net/. The sample plugins in XJC should help you get started on how to write one;
The jaxb2-commons project is intended to host plugins for JAXB 2;
For example, after writing to users@jaxb.dev.java.net, I was added as a developer and directories "fluent-api" and "www/fluent-api" were created. I can put whatever I want there. Files in "www/fluent-api" can be served from http://jaxb2-commons.dev.java.net/fluent-api/
Projects in java.net are encouraged to use "org.jvnet.", so I used "org.jvnet.jaxb2_commons" as my plugin's package prefix.
Feel free to update www/index.html with a link to www/fluent-api/
Pick the license of your choice.

Hsql File Locking failed in Ant + Windows

2006-01-30T17:23:00.000-08:00

I had two junit tests using hsql to perform db operations. They were run using Ant 1.6.5 with hsql-1.8.0.2. When run on Linux 7.2, they worked fine. When run in Eclipse 3.1.1 (without using Ant), they also worked fine.

However, when they were run in ant 1.6.5 + Cygwin on Windows, the 2nd test always failed saying it couldn't acquire the lock as another process already had it.

Thanks to Campbell Boucher-Burnet, the problem can be resolved via the shutdown=true connection property. Details can be found here.

More on hsql connection properties can be found here.

Comparing IDE's on support of Java 5 Generics

2006-01-29T12:10:00.000-08:00

One interesting thing I've observed from building jaxb 2.0 (from source) is that if I set up a project in Netbeans (4.1.x) or IntelliJ (5.0.x) or just build it directly via Ant (1.6.5), the source code compiles just fine. But if I set up a project in Eclipse (3.1.2), a lot of source codes with Java 5 generics constructs are marked as errors by the IDE! (Example: ElementInfoImpl.java under package com.sun.xml.bind.v2.model.impl)

This seems to indicate that as jar as Java generics goes Eclipse is not quite up to the task yet.

I am not sure if they have anonymous cvs access, but if you subscribe to www.dev.java.net (which is free), you can check out the jaxb 2.0 cvs source tree and see the IDE differences (simply by importing the existing project config files set up for various IDE's including Eclipse, Netbeans and IntelliJ):

cvs -d:pesrver:yourid@cvs.dev.java.net:/cvs co -d jaxb-ri jaxb2-sources/jaxb-ri

I filed a bug report to Eclipse https://bugs.eclipse.org/bugs/show_bug.cgi?id=125956.

Update on 23Feb06: I received a email from Kent Johnson saying he has this bug fixed in Eclipse 3.2M6. Thanks Kent!

JAXB 2.0 XJC Fluent API Plugin launched!

2006-01-26T22:50:00.000-08:00

As mentioned in my previous post, one thing I wish to have is the support of method chaining in JAXB. It turns out it's really easy to write a JAXB 2.0 XJC plugin. Indeed, you can now download it from https://jaxb2-commons.dev.java.net/fluent-api/.

It's also announced at the acquarium and mentioned here.

Special thanks to Kohsuke Kawaguchi.

Fibonacci Numbers in Scheme

2006-01-24T10:29:00.002-08:00

A Fibonacci number is basically the sum of the previous two.

pos: 0 1 2 3 4 5 6 7  8  9  10 ...
fib: 0 1 1 2 3 5 8 13 21 34 55 ...

Simple but slow version in Scheme:

; Time is of exponential order of N; whereas
; space is of linear order of N
(define (fibslow N)
 (cond ((< N 2) N)
       (else (+ (fibslow (- N 1))
                (fibslow (- N 2))))))

How can we make it faster ?

Warning: reading futher will reveal a faster version.

Fast version in Scheme:

; Time is of linear order of N; whereas
; space is of constant order of N
(define (fibfast N)
 (cond ((< N 2) N)
       (else (fibup N 2 1 0))))

(define (fibup MAX count n-1 n-2)
 (cond ((= MAX count) (+ n-1 n-2))
       (else (fibup MAX (+ 1 count) (+ n-1 n-2) n-1))))

To feel the performance difference, try:

(fibslow 30)

vs.

(fibfast 30)

Tower of Hanoi in Scheme

2006-01-23T22:02:00.000-08:00

;Solving the Tower of Hanoi in Scheme
(define (move N from to spare)
  (if (= N 0)
      (display "")
      (begin 
        (move (- N 1) from spare to)
        (display "move from ")(display from)(display " to ")(display to)(display "\n")
        (move (- N 1) spare to from))))

Sample run:

> (move 3 'start 'end 'spare)
move from start to end
move from start to spare
move from end to spare
move from start to end
move from spare to start
move from spare to end
move from start to end

How can we solve it iteratively (ie via tail recursion) ?

Warning: reading futher will reveal an iterative version.

It requires a lot more code and not in any way faster.

; Solving the Tower of Hanoi iteratively in Scheme
(define (moveIter N from to spare)
  (moveIterImpl N from to spare () ()))

(define (moveIterImpl N from to spare displayList moveList)
  (if (= N 0)
      (if (null? displayList)
         (display "")
         (begin
           (output (car displayList))
           (if (null? moveList)
               (display "")
               (eval (toMoveIterCmd moveList displayList)))))
      (moveIterImpl (- N 1) from spare to 
        (cons (toPair from to) displayList)
        (cons (toMoveParam (- N 1) spare to from) moveList))))

(define (toMoveIterCmd moveList displayList)
  (cons 'moveIterImpl 
        (cons (car (car moveList))
              (cons (cons 'quote (cons (car (cdr (car moveList))) ()))
                    (cons (cons 'quote (cons (car (cdr (cdr (car moveList)))) ()))
                          (cons (cons 'quote (cons (car (cdr (cdr (cdr (car moveList))))) ()))
                                (cons (cons 'quote (cons (cdr displayList) ()))
                                      (cons (cons 'quote (cons (cdr moveList) ()))
                                            ()))))))))

(define (toMoveParam N from to spare)
  (cons N (cons from (cons to (cons spare ())))))

(define (output displayItem)
  (begin
    (display "move from ")(display (car displayItem))
    (display " to ")(display (car (cdr displayItem)))
    (display "\n")))

(define (toPair from to)
  (cons from (cons to ())))

To run, simply type:

> (moveIter 3 'start 'end 'spare)

MIT Open Courseware

2006-01-23T19:39:00.000-08:00

http://ocw.mit.edu/index.html

Trivial Java Quiz

2006-01-21T21:03:00.000-08:00

Can you name a public method in JDK1.5 that exists in a class, but cannot be found declared or defined anywhere in the class (source file) ? In other words, you can invoke the method and it will behave as expected, but the method (and even the method name) cannot be found in the source code of the class (or it's super classes or super interfaces), nor in the respective javadoc.

Warning: reading futher will reveal the answer.

Here is one answer:

It appears there are two static methods generated for every enum class:

method values(), which returns the full list of enum instances of the specific Enum class; and
method valueOf(String), which returns the respective enum instance with the same name as the input string.

For (2), interestingly there is also a public static method valueOf with 2 parameters defined in the source file of Enum:

// existing method in Enum
public static <T extends Enum<T>> T valueOf(Class<T> enumType, String name);

For (1), it puzzles me why there isn't also a public static method like:

// why not
public static <T extends Enum<T>> T[] values(Class<T> enumType);

The Enum javadoc doesn't even mention the 2 generated static methods, although one can see them when using auto-completion in an IDE. Seems like an oversight to me.

Structure and Interpretation of Computer Programs

2006-01-21T09:09:00.000-08:00

http://swiss.csail.mit.edu/classes/6.001/abelson-sussman-lectures/

The video is pretty cool, and the online book is free!