Re: webkitgtk+

[Svante Signell <svante.signell@telia.com>]

On Mon, 2011-10-10 at 18:59 +0200, Svante Signell wrote:
> Ssmuel,
> 
> looks like webkitgtk+ build fine with one of the Sergios patches:
> 
> 2011-10-08:
> 18:01:04< gnu_srs> what about webkitgtk+? slpz made it build, but the
> buildds don't have his patches yet.
> 18:02:20< youpi> did he complete it?
> 18:02:28< youpi> I know he used it to stress-test
> 18:02:34< gnu_srs> I think so.
> 18:02:37< youpi> but I don't remember him saying he could finish
> 18:02:45< youpi> the issue is at the last stages, during dh_strip & such
> 18:06:28< gnu_srs> From 9 sep 2009: 00:09:37< slpz>  dpkg-source --after-build webkitgtk+-1.4.2
> 18:06:41< gnu_srs> s/2009/2011/g
> 18:08:32< gnu_srs> And Oct 3 2011: 10:58:24< slpz> gnu_srs: That patch
> is convenient, but not critical. I think the only patch that is really
> needed is the one for "mach_msg fails to copy out-of-line data when
> length >= 512MB" and perhaps the one which modifies external pageout (I
> haven't tried building webkitgtk+ without it, so I can't tell)
> 
> 2011-10-10:
> (05:34:46 PM) srs: slpz: Did your build of webkitgtk+ complete? I was asked by youpi to find out.
> (05:59:03 PM) slpz: gnu_srs: Sure, several times.
> (06:40:54 PM) antrik: slpz: did you build the whole Debian package?
> (06:41:23 PM) antrik: (that was the actual question -- on the buildds
> apparently it often fails not during compilation, but later on during
> the packaging stages...)
> (06:55:34 PM) slpz: antrik: Yes, with dpkg-buildpackage.

Looks like there maybe are yet unpublished patches needed to get
webkitgtk+ to build properly. (I have not tried to build webkitgtk+
again due to disk and memory space limitations.) Hoping for some luck
with coreutils and ruby1.9.1 too, which both hangs during the tests with
the latest gnumach (2011-10-10).

irc 2011-10-03:

10:48:45< gnu_srs> slpz: Did you succeed to build webkitgtk+ with the
"Sync pager before clearing MAY_CACHE flag patch" and/or with any of
your other patches?

10:58:24< slpz> gnu_srs: That patch is convenient, but not critical. I
think the only patch that is really needed is the one for "mach_msg
fails to copy out-of-line data when length >= 512MB" and perhaps the
one which modifies external pageout (I haven't tried building
webkitgtk+ without it, so I can't tell)

11:01:56< slpz> gnu_srs: The latter is bigger than I expected, so I
don't know if sending it to the mailing list would be convenient.
Perhaps publishing an alternative git branch/repo with those changes
would be better, to have them well tested before proposing for formal
inclusion.

11:04:45< gnu_srs> Since webkitgtk+ and other packages never completes
it would be nice to get patched gnumach/hurd into the buildds asap.

11:06:06< gnu_srs> In any case, when you have published these patches I
can try them out until Samuel publish new versions.