W3C home > Mailing lists > Public > public-cwm-talk@w3.org > July to September 2004

Re: a profiler for cwm

From: Dominique HazaŽl-Massieux <dom@w3.org>
Date: Wed, 28 Jul 2004 17:37:57 +0200
To: Yosi Scharf <syosi@MIT.EDU>
Cc: Dan Connolly <connolly@w3.org>, public-cwm-talk@w3.org, Vivien Lacourba <vivien@w3.org>
Message-Id: <1091029077.6022.456.camel@stratustier>
Hello Yosi,

Le mer 28/07/2004 ŗ 17:19, Dan Connolly a ťcrit :
> The W3C tech reports page is maintained using cwm.
>   http://www.w3.org/2002/01/tr-automation/
> I reproduced the relevant files under
>   http://www.w3.org/2000/10/swap/test/tr-aut/
> if you go there and do
> 	$ make tr.rdf
> you can watch cwm run for a loooong time. I'm sure
> Vivien would appreciate it if you could make it
> go faster.

In case that helps, there are more details at:
(of course, the data there are outdated); in particular, there was a
serious performance regression between the (very old) version of cwm
tagged as "fast_for_dom", and the version I checked at that time.

We're using this "fast_for_dom" version in production, since that's the
only way of making TR automation usable. Being able to use newer
versions of cwm would be a very good thing.

(FWIW, the revisions 1.2 of filter-trdata.n3 and tr-process.n3 includes
fixes on the 1.1 ones that are not in the production set up, since they
don't work properly in the old version of cwm we're using)

Feel free to ask if you have any question or suggestion on the way we
use cwm in this project.

Dominique HazaŽl-Massieux - http://www.w3.org/People/Dom/

Received on Wednesday, 28 July 2004 11:40:18 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:01:04 UTC