ingo schuster wrote:
>
> At 13:01 2000-12-15, [EMAIL PROTECTED] wrote:
> >This is mainly for David and Santiago:
> >
> >Do you plan to commit updated features for the profiler
> >or the diskcache in order to incorporate it into 1.3a1 ?
> >
> >If no, I'll feature freeze 1.3A1 aned fork out a 1.3a1
> >release branch around 19:00 GMT.
>
> Ok, I've finally committed all features that I liked to see in 1.3a1.
> Feature freeze is ok with me.
>
> > From then on, all the committers/users willing to participate
> >in the the clean-up and bug fix process should checkout and
> >work on the release branch rather than the HEAD.
>
> Huh, I never committed using branches, but I hope WinCVS offers support for
> dummies. :-)
Branch tags are "sticky", so once you do update -r branch your files get
commited against this branch until you do update -a. What looks
difficult for me is merging, that will definitely require lots of tests
and manual reading to make sure. I did it once, but I had to solve a lot
of conflicts, because I think I did not do it in the proper way.
To avoid problems when I don't feel sure about what I'm doing with cvs,
I use to check in file by file (or at least a leaf directory at a time)
and do a "diff" first, just to know what I'm going to spoil in advance
:-). Also, it allows more focused comments if changes cover several
issues (some bugs fixed and also some change).
>
> ingo.
>
> >The HEAD will stay open during this process so new features can
> >be added to Jetspeed during this time, they just won't be
> >included in 1.3a1
I have been looking to the remaining multithreading problems with Juan
Carlos. It does not look difficult to fix, and it is important to get it
right ASAP. He is working in this area and we think we should open a
branch for this work soon. I was waiting for my commit of the cache, but
if it is not in by monday, I think it would be better to start it in
parallel. I don't know if I'll have it tested and working by then.
We discovered that when complex PSML is used, a lot of time is wasted by
regenerating recursively the PortletSet structure. This process is not
really needed, unless the PSML changes.
Also, the controllers and controls store state inside, somethings that
breaks under load, and brings unefficiencies and garbage to the process.
As a result of our tests, PanedPortletControl(ller) breaks when several
requests are running at the same time against the same page (in my
setup, it starts breaking at about 10 parallel requests).
> >
> >--
> >Rapha�l Luta - [EMAIL PROTECTED]
> >Vivendi Universal Networks - Services Manager / Paris
> >
> >
> >--
> >--------------------------------------------------------------
> >Please read the FAQ! <http://java.apache.org/faq/>
> >To subscribe: [EMAIL PROTECTED]
> >To unsubscribe: [EMAIL PROTECTED]
> >Archives and Other: <http://marc.theaimsgroup.com/?l=jetspeed>
> >Problems?: [EMAIL PROTECTED]
>
> --
> --------------------------------------------------------------
> Please read the FAQ! <http://java.apache.org/faq/>
> To subscribe: [EMAIL PROTECTED]
> To unsubscribe: [EMAIL PROTECTED]
> Archives and Other: <http://marc.theaimsgroup.com/?l=jetspeed>
> Problems?: [EMAIL PROTECTED]
--
--------------------------------------------------------------
Please read the FAQ! <http://java.apache.org/faq/>
To subscribe: [EMAIL PROTECTED]
To unsubscribe: [EMAIL PROTECTED]
Archives and Other: <http://marc.theaimsgroup.com/?l=jetspeed>
Problems?: [EMAIL PROTECTED]
