On Thu, 12 Dec 2002, Nicola Ken Barozzi wrote:
> > Bruno Dumon wrote: > > On Thu, 2002-12-12 at 10:44, Steven Noels wrote: > > > >>Christian Haul wrote: > >> > >> > >>>On 12.Dec.2002 -- 10:08 AM, [EMAIL PROTECTED] wrote: > >>> > >>> > >>>>Anyway, whatever the outcome I look forward to this new development as it > >>>>certainly is something that is needed badly. Perhaps Gianugo and Nicola > >>>>could provide some details on what they are thinking of? > >>> > >>> > >>>What we were discussing is a possibility to debug the pipeline. For > >>>this we have currently the LogTransformer. One could put this between > >>>several transformation steps and see what happened. > >> > >>sounds very similar to captor which has been posted once as a patch by > >>Bruno: http://outerthought.net/captor.html > > > > well, there's a difference though. Captor is very useful in itself for > > people wanting to understand pipelines, though its completely un-useful > > if you're writing your own SAX-transformer which (during development) > > might be quite buggy and generate invalid SAX-events. But for many users > > out there captor can be extremely useful. > > Our idea is based on the profiling component in Cocoon CVS, so yes, we > have seen it. I have included the concept the 'Captor' in the profiling pipelines for some time ago, but only for Cocoon 2.1.(Thanks to Bruno) > Anyway, how that tool works is flawed, and will not necessarily work > with all pipelines (this came out in the Stammtisch), since they might > rely on the events being processed through the whole pipeline instead of > being cached. Do you mean the 'Captor' component or the profiling component? To allow debuging on a server, you should only need to 'instrumentable' the profiling component, I think. My 2 cents, Stephan Michels. --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, email: [EMAIL PROTECTED]
