[Kde-perl] PerlQt-3.004 Final

[Germain Garand]

Here it is,
PerlQt-3.004 Final is available at http://www.sourceforge.net/projects/pe=
rlqt

Thanks to all the people who gave feedback about the RCs : you are awesom=
e !

ChangeLog :

2002-12-16 GG (v.3.004-final)
         - test suite ("make test" target)=20
         - one more example (progress.pl)
         - fixed a bug related to pointer mapping and Smoke's cf_virtual =
flag
           (regeneration of Smoke isn't mandatory, but is recommended)
         - fixed possible redefinition of a sub in isa.pm
         - raised default test threshold to 10
         - some code optimizations (object construction/destruction)
         - added an undocumented but supported CAST function
           e.g  CAST $obj, "Qt::Application"; (it proved to be useful in =
some
           situations)

And as a shameless incitation to upgrade ;) here are some comparative=20
benchmarks :

PerlQt-3.002 :

Benchmark started on lun d=E9c 2 03:35:05 2002
Timing 20000 method calls...878ms
Timing 20000 static enum calls (no parenthesis)...668ms
Timing 20000 static calls (with parenthesis)...728ms
Timing 20000 object creations...3247ms
Timing 20000 object deletions...1126ms
Timing 10 ambiguous method calls (without cache)...11ms
Timing 10 ambiguous method calls...3ms

PerlQt-3.004 :

Benchmark started on lun d=E9c 16 18:02:36 2002
Timing 20000 method calls...575ms
Timing 20000 static enum calls (no parenthesis)...730ms
Timing 20000 static calls (with parenthesis)...774ms
Timing 20000 object creations...2584ms
Timing 20000 object deletions...696ms
Timing 10 ambiguous method calls (without cache)...10ms
Timing 10 ambiguous method calls...2ms

So except static calls that have slightly suffered from the debugging=20
overhead, it's an overall 30 to 40% gain (yay! even PyQt and its early=20
binding method is now lagging behind!).

G.