Java parser update: thesis roundup

[Roberto Raggi]

Hi Jakob!

On Wednesday 01 February 2006 17:20, Jakob Petsovits wrote:
> I'll see what I can do. Is your preferred email address the kdevelop.org
> one or should I use some other one for contacting you?

It is very difficult to me to read my email at kdevelop.org... a lot of 
spam :-) please try to use my address @trolltech even better use the 
kdevelop-devel mailing list. I'm sure a lot of users are interested in your 
work! and it's very good for us to involve more developers.

>
> > Jakob, how much slower(faster?) is kdev-pg compared to other Java
> > parsers? btw I saw you use flex to generate the scanner. I know the
> > generated flex scanner is slower than most of the other scanner
> > generators (antlr, re2c, ...).

I did some testing with your parser. Well.. it is *really* fast :-)

Project GNU classpath

[rraggi at ... classpath]$ find -name "*.java" | wc -l
4175

[rraggi at ... classpath]$ find -name "*.java" -exec cat {} \; | wc -l
955070

- KDevelop Java (compiled with -O2 -s)
[rraggi at ... classpath]$ time find -name "*.java" -exec ~/lib/bin/kdev-java {} 
\; -print >& /dev/null

real    0m33.784s
user    0m19.349s
sys     0m14.045s


- ANTLR C++ (compiled with -O2 -s)
[rraggi at ... classpath]$ time find -name "*.java" -exec ~/lib/bin/antlr-java {} 
\; -print >& /dev/null

real    0m46.130s
user    0m31.038s
sys     0m14.985s

> Thanks to you, without kdevelop-pg that thesis would not have been possible
> in this form. But enough of mutual acknowledgements, I'll get back to work
> (or whatever).
:-)

ciao robe