statcvs-users — StatCVS Users Mailing List

[Statcvs-users] statcvs: exclution

[Nereson M. R S. 8. CSPTS/S. <Mic...@sc...>]

Is there a way to exclude directories or files that statcvs reports on?

=20

For example, under my project root, I have my prototype files and my
source files. I don't want any of the prototype files to be considered
in the statcvs reports.

=20

Thanks for any help that can be provided.

=20

~ Mike

Re: [Statcvs-users] statcvs report does not show all commit comments.

[Richard C. <ri...@cy...>]

Hi Hussein,

Am 01.03.2005 um 20:52 schrieb Hussein Vastani:
> I noticed that statcvs does not generate the complete commit comments 
> history for a given cvs resource/author. For instance at the bottom of 
> some of the reports there's a "(18 more)" label but its not a link 
> that would lead to the remaining 18 commit comments for that 
> resource/author. That would certainly be useful.

Yes, this has been requested several times now. I guess we should show 
the author's whole log.

> Is it configurable?

No, not at the moment.

The issue here is this: An author can have hundreds or thousands of 
commits. Showing them all on the autor's page would make the page huge. 
We could paginate the author commit log (like we do for the main commit 
log), but this seems a bit ugly to me because it could lead to so many 
files. Anyway, I guess that's the way to go.

Any other ideas?

Richard

Re: [Statcvs-users] unexpected date format in line 17

[Jukka U. <juk...@dn...>]

Ignacio Canals wrote:
> Hi,
> 	I have installed on the Server
> Kaffe Virtual Machine
> 
> Engine: Just-in-time v3   Version: 1.1.x-cvs   Java Version: 1.1
> Configuration/Compilation options:
>   Compile date     : Sun Dec 19 12:19:51 UTC 2004
>   Compile host     : cyberhq
>   Install prefix   : /usr/lib/kaffe/pthreads
>   Thread system    : unix-pthreads
>   Garbage Collector: kaffe-gc
>   CC               : /usr/bin/gcc-3.4
> 
> I haven't checkout the latest version of statcvs and build it because I
> don't have an account in cvs.sourceforge.net. However, that should be the
> same than download the latest compiled version, or not?

You don't need sourceforge account for checkout. You can use anonymous
account. https://sourceforge.net/cvs/?group_id=57558

But do you tried this one?
http://page.mi.fu-berlin.de/~cyganiak/temp/statcvs-0.2.2.zip

If this does not help, there is new problem with date format.
Maybe it would be good idea to make possible to pass date format
from statcvs command line?

- Jukka -

[Statcvs-users] statcvs report does not show all commit comments.

[Hussein V. <va...@ac...>]

Hi,
Congratulations on a very useful project and thankyou for you for your 
wonderful work!

I noticed that statcvs does not generate the complete commit comments 
history for a given cvs resource/author. For instance at the bottom of 
some of the reports there's a "(18 more)" label but its not a link that 
would lead to the remaining 18 commit comments for that resource/author. 
That would certainly be useful. Is it configurable?

regards:

Hussein Vastani,
Actenum Corp,
Vancouver, Canada.

Re: [Statcvs-users] unexpected date format in line 17

[Jukka U. <juk...@dn...>]

Ignacio Canals wrote:
> Jukka,
> 	Hi. Thanks for your answer. I download the 0.2.2 release from the
> website and I still have the same problem. Exactly the same error in the
> same line. Do you have any idea? What could I do?
> Thanks for your answer,
> Ignacio Canals
> 
> 

Hi,

I think there are one fix made for log file timestamp format after
0.2.2 release.

Look 
http://cvs.sourceforge.net/viewcvs.py/statcvs/statcvs/src/net/sf/statcvs/input/CvsRevisionParser.java?rev=1.40&view=log

Do you have tried to checkout latest source files from statcvs 
repository and build statcvs? I think that should solve problem.

There has some discussion about same issue before

https://sourceforge.net/mailarchive/forum.php?thread_id=5760001&forum_id=11441

- Jukka -


> -----Original Message-----
> From: Jukka Uusisalo [mailto:juk...@dn...] 
> Sent: Viernes, 25 de Febrero de 2005 1:46
> To: Ignacio Canals
> Subject: Re: [Statcvs-users] unexpected date format in line 17
> 
> Ignacio Canals wrote:
> 
>>Hi,
>>
>>            First than all I want to congratulate you for statcvs. Is 
>>really a great idea, specially to measure performance of  developers. 
>>However, I install the application on my server but I am still not able 
>>to make it work. It throws me this message:
>>
>> 
>>
>>unexpected date format in line 17ld
>>
>> 
>>
>>and the line 17 is:
>>
>>date: 2005-02-23 22:28:40 +0000;  author: ignacio;  state: Exp;
>>
>> 
>>
>>Do you know what can I do to fix it?
>>
>>Thanks in advance for the help and I really hope you will continue 
>>developing this useful tool.
>>
> 
> 
> Hi,
> 
> Do you have 0.2.2 release?
> 
> There are at least one fix for log file timestamp format after 0.2.2 
> release, so if build statcvs from source files from statcvs 
> cvs-repository this should be fixed.
> 
> - Jukka -
> 
> 

[Statcvs-users] unexpected date format in line 17

[Ignacio C. <ic...@le...>]

Hi,

            First than all I want to congratulate you for statcvs. Is really
a great idea, specially to measure performance of  developers. However, I
install the application on my server but I am still not able to make it
work. It throws me this message:

 

unexpected date format in line 17

 

and the line 17 is:

date: 2005-02-23 22:28:40 +0000;  author: ignacio;  state: Exp;

 

Do you know what can I do to fix it?

Thanks in advance for the help and I really hope you will continue
developing this useful tool.

Greetings,

Ignacio Canals

Re: [Statcvs-users] A new article on StatCVS...

[Tom C. <to...@in...>]

On Wed, 2005-02-23 at 23:37 +0100, Richard Cyganiak wrote:
> Hi Tom,
> 
> Am 23.02.2005 um 16:50 schrieb Tom Copeland:
> > ...by yours truly is right here:
> >
> > http://www-128.ibm.com/developerworks/java/library/j-statcvs/
> >
> > Thanks again for a fine piece of software!
> 
> Thanks, that's a great article. Needless to say I enjoyed reading it :-)

Quite right!  And I've enjoyed working with StatCVS, so it's a fair
trade :-)

> I've added a Further Reading section to the manual, with a link to your 
> article.

Thanks much!

Yours,

Tom

Re: [Statcvs-users] A new article on StatCVS...

[Richard C. <ri...@cy...>]

Hi Tom,

Am 23.02.2005 um 16:50 schrieb Tom Copeland:
> ...by yours truly is right here:
>
> http://www-128.ibm.com/developerworks/java/library/j-statcvs/
>
> Thanks again for a fine piece of software!

Thanks, that's a great article. Needless to say I enjoyed reading it :-)

I've added a Further Reading section to the manual, with a link to your 
article.

Richard

[Statcvs-users] A new article on StatCVS...

[Tom C. <to...@in...>]

...by yours truly is right here:

http://www-128.ibm.com/developerworks/java/library/j-statcvs/

Thanks again for a fine piece of software!

Yours,

Tom

[Statcvs-users] RE: question on generation

[David S G. <Dav...@la...>]

Aaron,

It sounds like what you really want is the 'cvs annotate' feature.  This cvs
command prints out the revision number and last editor for each line of each
file in your working CVS directory.  If you check out an older revision of a
file, you will see the 'history' of that file's lines up to only the version
you have checked out.

I don't know how statcvs does the line (was curious myself) counting, but
`cvs annotate | grep -c "username"` does a good job of counting lines of a
file last edited by 'username'.  If you add a pipe in there before the grep
with a `grep ';'`, you'd probably get a reasonable estimate of lines of
actual code.

DSG

> To: sta...@li...
> From: Aaron Zinman <az...@MI...>
> Date: Mon, 21 Feb 2005 22:35:58 -0500
> Subject: [Statcvs-users] question on generation
> 
> Hi, I'm currently working on a research project that is a 
> nice visualization that shows the relationship between code 
> and communication in open source projects.  Now that I have 
> my visualization built, I need to get real data.  I'm new to 
> dealing with CVS in terms of analysis, and uour project 
> seemed like it would help a lot.  I basically need to know 
> who owns what lines of code at what point in time.  However, 
> I'm not quite sure how you get the statistics that you do 
> only from the "cvs log" command.  Your generated reports 
> indicate that you can know that "cyganiak" has 4323 lines of 
> code, but how are you getting this information?  That is, cvs 
> logs give you
> 
> revision 1.2
> date: 2003/03/25 01:20:13;  author: lukasz;  state: Exp;  
> lines: +20 -2 Minor code review tasks.
> Added some javadocs
> Fixed some unhandled Exceptions
> 
> 
> So you know that lukasz added 20 lines of code, but you don't 
> know who's -2 he got rid of, so how could you properly deduct 
> them from the right other people?
> 
> Any pointers would be appreciated.
> 
> Thanks,
> Aaron

Re: [Statcvs-users] question on generation

[Richard C. <ri...@cy...>]

Hi Aaron,

sounds like an interesting research project. Is there anything to look 
at yet?

Am 22.02.2005 um 04:35 schrieb Aaron Zinman:

> I basically need to know who owns what lines of code at what point in 
> time.  However, I'm not quite sure how you get the statistics that you 
> do only from the "cvs log" command.  Your generated reports indicate 
> that you can know that "cyganiak" has 4323 lines of code, but how are 
> you getting this information?  That is, cvs logs give you
>
> revision 1.2
> date: 2003/03/25 01:20:13;  author: lukasz;  state: Exp;  lines: +20 -2
> Minor code review tasks.
> Added some javadocs
> Fixed some unhandled Exceptions
>
> So you know that lukasz added 20 lines of code, but you don't know 
> who's -2 he got rid of, so how could you properly deduct them from the 
> right other people?

We don't deduct them. You're referring to the chart and table on the 
Authors page of the StatCVS report. The story we want to tell there is: 
How much did each author, over time, contribute to the project? We 
don't want to show who has written how much of the code that is 
up-to-date *right now*, but how much they've written *over time*. So 
it's reasonable to count 20 lines for lukasz, and the person who wrote 
the original 2 lines keeps them. The contribution doesn't become 
worthless just because it was later replaced by something else.

This probably doesn't help you much.

But here's something that you could use: the "cvs annotate" command. We 
don't use the data from that, but you probably want to.

In a checked-out working copy, run: "cvs annotate SomeFile.java". This 
gives you a breakdown of who owns which line of that file *right now*. 
"cvs annotate -r 1.5 SomeFile.java" shows you who owned which line at 
the time of revision 1.5 of the file. From "cvs log", you can get the 
date of revision 1.5. I guess this is exactly the data you need.

Richard

[Statcvs-users] question on generation

[Aaron Z. <az...@MI...>]

Hi, I'm currently working on a research project that is a nice 
visualization that shows the relationship between code and 
communication in open source projects.  Now that I have my 
visualization built, I need to get real data.  I'm new to dealing with 
CVS in terms of analysis, and uour project seemed like it would help a 
lot.  I basically need to know who owns what lines of code at what 
point in time.  However, I'm not quite sure how you get the statistics 
that you do only from the "cvs log" command.  Your generated reports 
indicate that you can know that "cyganiak" has 4323 lines of code, but 
how are you getting this information?  That is, cvs logs give you

revision 1.2
date: 2003/03/25 01:20:13;  author: lukasz;  state: Exp;  lines: +20 -2
Minor code review tasks.
Added some javadocs
Fixed some unhandled Exceptions


So you know that lukasz added 20 lines of code, but you don't know 
who's -2 he got rid of, so how could you properly deduct them from the 
right other people?

Any pointers would be appreciated.

Thanks,
Aaron

Re: [Statcvs-users] Questions & Thanks

[Robin G. <rg...@bl...>]

>Hi,
>
>Do you still have this file in your repository? I just upload patches to 
>fix this one and if you still have possibility to test this issue, i can 
>send you statcvs.jar if you don't want build statcvs.jar yourselves with 
>patch files.
>
>- Jukka -

Yes - if you could send me the jar, it would be great - I have not had the 
time to set up things to compile from cvs yet.

-Robin 

Re: [Statcvs-users] Questions & Thanks

[Jukka U. <juk...@dn...>]

Robin Getz wrote:
> 
> 2) I found a "bad" file in one of the repositories named '*?'. this Made 
> the log something like:
> 
> ---snip----
> RCS file: /cvsroot/uboot533/u-boot_1.1.1/disk/*^M,v
> Working file: u-boot_1.1.1/disk/*^M
> head: 1.1
> branch: 1.1.1
> ---snip----
> 
> On this, statcvs complains:
> ----------
> 
> Logfile parsing failed.
> line 58478: expected 'Working file: ' but found ',v'
> 
> -----------
> 
> Why the file is there - I don't know (and will be removing it soon) but 
> wouldn't it be better to create the rest of the stats without one file, 
> than stop working all together - if only one filename is bad?
> 

Hi,

Do you still have this file in your repository? I just upload patches to 
fix this one and if you still have possibility to test this issue, i can
send you statcvs.jar if you don't want build statcvs.jar yourselves with
patch files.

- Jukka -

Re: [Statcvs-users] error saying that logfile empty

[Brian G. P. <br...@br...>]

On Monday 24 January 2005 07:26 am, Satish Babu G wrote:
> Hi this is satish. I got a problem when I was trying to run the statcvs
> command it is saying that the logfile is empty. I just added one text file
> into the cvs directory with ext .log and then commit to the cvs. After I
> tried to run the statcvs command in the command prompt. It is failing and
> given a error saying that empty log file. Could u plz tell me how to solve
> this one. Waiting for u r reply.

The cvs log file is genrated by the 'cvs log' command.  See my other message 
for a script that should get the log for you.  

Please also review the statcvs documentation here:
http://statcvs.sourceforge.net/

Also, for many users of this list, and many developers of StatCVS, English is 
not their native language (nor I suspect your own native language).  Please 
follow common Internet forum/list etiquette and use proper grammar and 
spelling. u r l33t 5p33k bugs me.

Regards,

  - Brian

[Statcvs-users] error saying that logfile empty

[Satish B. G <sat...@cg...>]

Hi this is satish. I got a problem when I was trying to run the statcvs
command it is saying that the logfile is empty. I just added one text file
into the cvs directory with ext .log and then commit to the cvs. After I
tried to run the statcvs command in the command prompt. It is failing and
given a error saying that empty log file. Could u plz tell me how to solve
this one. Waiting for u r reply.

Ok. 

Regards,

G.Satish.

 

 

Re: [Statcvs-users] connecting to remote repositary using statcvs

[Brian G. P. <br...@br...>]

On Monday 24 January 2005 03:48 am, Satish Babu G wrote:
> Hi. This is satish from CGI BANGALORE India. I want to know how to connect
> remote repository using statcvs. My requirement is like this. I need to
> connect to sourceforge London using pserver protocol. Is there any way (at
> least through ant scripting ) to connect to remote repositary through
> statcvs. Please let me know how to connect. Waiting for your reply.

You need a full cvs log.  Read the documentation, all the details are there.
PServer can be painfully slow to generate the full log on a large project.

For a quick overview, here are the steps that I use for a Sourceforge project 
that I am the administrator of (modified to use pserver):

#!/bin/bash

HOME=/home/brian/cvs

cd $HOME

echo "Updating XRMS CVS"

cvs -q -z3 -d :pserver:ano...@cv...:/cvsroot/xrms checkout 
xrms

echo "generating log"

cd xrms

cvs log > cvslog

cd $HOME

echo "generating statistics"

cd xrmsstats

/usr/java/bin/java -jar statcvs.jar -title "XRMS CRM" -verbose -viewcvs 
http://cvs.sourceforge.net/viewcvs.py/xrms/xrms/ ../xrms/cvslog ../xrms/

I've separated each single line of the shell script above with a blank line to 
account for inevitable email wrapping.

The approach above can be incredibly slow via PServer.  It may be faster to 
make a complete local copy of the entire repository. Sourceforge creates a 
daily backup tarball of the entire CVS repository:

wget http://cvs.sourceforge.net/cvstarballs/xrms-cvsroot.tar.bz2
wget http://cvs.sourceforge.net/cvstarballs/projectname-cvsroot.tar.bz2

I hope this helps.

Regards,

  - Brian

[Statcvs-users] connecting to remote repositary using statcvs

[Satish B. G <sat...@cg...>]

Hi. This is satish from CGI BANGALORE India. I want to know how to connect
remote repository using statcvs. My requirement is like this. I need to
connect to sourceforge London using pserver protocol. Is there any way (at
least through ant scripting ) to connect to remote repositary through
statcvs. Please let me know how to connect. Waiting for your reply.

Ok.

Regards,

G.Satish.

Re: [Statcvs-users] Symbolic Names patch

[Richard C. <ri...@cy...>]

Am 11.01.2005 um 23:15 schrieb Steffen Pingel:

> Hi,
>
>> I tested this patch and support for tag names is pretty critical
>> requirement. In my case, there was automated build system and if
>> there was any commits in cvs, there was new build and tag with build
>> number every day. Repository was three years old, so graphics was
>> unreadable.
>
> Yes, I would definitly like to add a setting that allows for filtering 
> of tags
> by regular expression (StatCvs-XML has it and it works fine for me).
>
> Richard, what do you think?

Happy hacking :-)

Thanks to you guys, and sorry about my lack of involvement recently. 
Not enough time.

Cheers,
Richard

Re: [Statcvs-users] Symbolic Names patch

[Steffen P. <ste...@gm...>]

Hi,

> I tested this patch and support for tag names is pretty critical
> requirement. In my case, there was automated build system and if
> there was any commits in cvs, there was new build and tag with build
> number every day. Repository was three years old, so graphics was
> unreadable.

Yes, I would definitly like to add a setting that allows for filtering of tags 
by regular expression (StatCvs-XML has it and it works fine for me).

Richard, what do you think?

Steffen

-- 
Steffen Pingel - ste...@gm... - http://steffenpingel.de

Re: [Statcvs-users] Symbolic Names patch

[Jukka U. <juk...@dn...>]

Steffen Pingel wrote:
> Hi,
> 
> I merged the patch that adds support for symbolic names and parsing of 
> CVS/Entries files to check the working revision for loc counting. I 
> tagged the repository before applying the patch 
> (symbolic_names-1-before-merge) and after merging it ( 
> symbolic_names-1-merge). That way the delta can be removed easily if 
> needed.
> 
> Currently the symbolic names are displayed in the global and the authors 
> loc chart. It might prove useful to add support for filtering symbolic 
> names (tags) through regular expressions like StatCvs-XML has. Usually 
> it is not desired to have all tags in the chart (i.e. branch/patch merge 
> tags). Some people might even want to turn tags completly.
> 

I tested this patch and support for tag names is pretty critical
requirement. In my case, there was automated build system and if
there was any commits in cvs, there was new build and tag with build
number every day. Repository was three years old, so graphics was
unreadable.


> Please let me know, if there are any problems with the patch.
> 
> Steffen
> 
> 
> PS: Richard, could you take a look at this bug report: 
> http://sourceforge.net/tracker/index.php?func=detail&aid=1058363&group_id=57558&atid=484569 
> 

Yes, I think this is resolved. :)


BTW. Why statCVS and statCVS-XML are separated projects?


- Jukka -

Re: [Statcvs-users] Symbolic Names patch

[Jukka U. <juk...@dn...>]

Jukka Uusisalo wrote:
> Steffen Pingel wrote:
> 
>> Hi,
>>
>> I merged the patch that adds support for symbolic names and parsing of 
>> CVS/Entries files to check the working revision for loc counting. I 
>> tagged the repository before applying the patch 
>> (symbolic_names-1-before-merge) and after merging it ( 
>> symbolic_names-1-merge). That way the delta can be removed easily if 
>> needed.
>>
>> Currently the symbolic names are displayed in the global and the 
>> authors loc chart. It might prove useful to add support for filtering 
>> symbolic names (tags) through regular expressions like StatCvs-XML 
>> has. Usually it is not desired to have all tags in the chart (i.e. 
>> branch/patch merge tags). Some people might even want to turn tags 
>> completly.
>>
>> Please let me know, if there are any problems with the patch.
>>
> 
> I checked out latest and seems there are something missing in test-src.
> I got classes in src to compled but with test-src, i got lot of errors, 
> missing classes, missing methods etc.
> 

Sorry, that was my own mistake. I made new checkout and ant build.xml 
works fine but under eclipse, not from command line. And reason was 
'old' statcvs.jar in ant/lib directory. These ant/lib/*.jar file were 
added to compile class path, so test classes get never compiled against 
classes in build directory.

After very short test this patch seems to be ok, but tomorrow i'll run
this version of statcvs against real cvs reporsitories. I'll let you 
know what happened.

- Jukka -

Re: [Statcvs-users] Questions & Thanks

[Richard C. <ri...@cy...>]

Am 03.01.2005 um 08:09 schrieb Robin Getz:

> Richard Cyganiak wrote:
>> StatCVS was not written with performance in mind, and probably never 
>> will be.
>
> What about accepting partial logs, and "adding" it to the existing 
> dataset, and re-generating the pngs from that?
>
> For example - if statcvs is going to be run every night via cron - why 
> not just pass it one days worth of changes, and it would add this 
> information to the pre-existing html - and would not touch the files 
> which don't need to be changed.

Good suggestion, but your idea would probably not save quite as much 
time as you might think.

StatCVS spends most CPU time generating the images. Most images are 
timelines, and those should be regenerated as often as possible so that 
they all show the same date range, up to the day of report generation.

Working with partial logs would cut down on bandwith usage though, 
which also is a bottleneck in some cases.

So if anybody wants to take this up ...

Cheers,
Richard

RE: [Statcvs-users] Questions & Thanks

[Robin G. <rg...@bl...>]

Richard Cyganiak wrote:
> > I also noticed that while running on these large logs, it consumes
> > _very_ large amounts of CPU resources (while running, it takes 96-99%
> > of available CPU). Other than 'nice' any suggestions?
>
>No, I'm sorry. StatCVS was not written with performance in mind, and 
>probably never will be.

What about accepting partial logs, and "adding" it to the existing dataset, 
and re-generating the pngs from that?

For example - if statcvs is going to be run every night via cron - why not 
just pass it one days worth of changes, and it would add this information 
to the pre-existing html - and would not touch the files which don't need 
to be changed.

Is that something that might be considered?

Thanks
-Robin 

Re: [Statcvs-users] Questions & Thanks

[Richard C. <ri...@cy...>]

Robin,

Am 02.01.2005 um 18:45 schrieb Robin Getz:
> Also, when I was trying -cvsweb option, I passed it:
>  -cvsweb http://cvs.blackfin.uclinux.org/cgi-bin/cvsweb.cgi
>
> And statcvs passed me back urls formed like:
> http://cvs.blackfin.uclinux.org/cgi-bin/cvsweb.cgi/u-boot_1.1.1/cpu/ 
> bf533/interrupts.c.diff?r1=1.15&r2=1.16&f=h
>
> However, it needs to have a cvsroot= in it, like so:
> http://cvs.blackfin.uclinux.org/cgi-bin/cvsweb.cgi/u-boot_1.1.1/cpu/ 
> bf533/interrupts.c.diff?cvsroot=uboot533&r1=1.15&r2=1.16&f=h

This was recently fixed for ViewCVS. I just applied the same changes to  
the CVSWeb code. You have to add the ?cvsroot parameter to the -cvsweb  
URL, then the right URLs should be formed.

Can you check out the latest StatCVS from our CVS and compile and check  
if the URLs work now?


>> > 3) How are the user statistics -> modules -> pie chart created? What
>> > signifies when something should be kept together? or split apart?
>>
>> I'd have to look at the code to say for sure. If I remember correctly,
>> it works like this: By default, everything is kept separately. If all
>> of a directory's subdirectories taken together are less than some
>> threshold, then they will be lumped together. The intention was to
>> avoid lots of tiny pie slices, but it doesn't work very well, and a
>> case could be made for dropping these charts from the report
>> altogether.
>
> I actually think this is a great idea, but could be sorted by  
> sub-directory/file and things are added until they hit some threshold.  
> What happens on my logs is that 80% is grouped to "other".
>
> Check:
> http://statcvs.blackfin.uclinux.org/uboot533/user_lgsoft.html
> http://statcvs.blackfin.uclinux.org/gcc3/user_bernds.html

Yeah, that's why I said it doesn't work very well. Sounds like your  
approach might work better.

Again, could you submit a short RFE to the tracker so the idea doesn't  
get lost?


> We are doing a linux kernel port to the Blackfin processor, and our  
> log for the kernel cvs is 65Meg. Maybe some note about the -mx java  
> option in the doc would be nice?

I agree. Will do.


> I also noticed that while running on these large logs, it consumes  
> _very_ large amounts of CPU resources (while running, it takes 96-99%  
> of available CPU). Other than 'nice' any suggestions?

No, I'm sorry. StatCVS was not written with performance in mind, and  
probably never will be.

Cheers,
Richard