[Development] Backporting QTextLayout Performance Improvements to Qt4
Milian Wolff
mail at milianw.de
Mon Nov 18 11:49:13 CET 2013
On Monday 18 November 2013 10:11:40 Knoll Lars wrote:
> Yes, once the patch is in 5.x and it affects 4.8 as well, it¹s usually ok
> to back port.
I ask b/c back then Eskil wrote:
> Eskil Abrahamsen Blomfeldt Sep 5, 2012
> Patch Set 4: Do not submit
> I don't think these P3 optimizations should be back-ported to a patch
> release. Patch releases exist only to increase stability, and that's why we
> have the commit policy of only allowing critical fixes (regressions,
> crashes, blatant drawing errors, security). There's already some indication
> that an earlier optimization for this same bug has introduced a crash which
> is triggered by Qt Creator (see QTBUG-27140). I'm working to verify it now,
> but please be conservative when committing to a patch release branch, since
> that's the only way to avoid destabilization.
So just to clarify once more, backporting the changes now from Qt5 is OK? I'd
be OK with only backporting the changes from Qt 5.1 or a similar stable
branch.
Thanks
> On 18.11.13 11:06, "Milian Wolff" <mail at milianw.de> wrote:
> >Hey all,
> >
> >I'd like to ask again whether I could backport some performance
> >improvements
> >in QTextLayout/QTextEngine from Qt5 to Qt4.
> >
> >Most notably there is https://codereview.qt-project.org/#change,34112
> >which
> >effects us in Kate/KDE as outlined here:
> >https://bugs.kde.org/show_bug.cgi?id=225228
> >
> >The changeset above was merged in Qt5 and matured there, see
> >https://codereview.qt-project.org/#change,34118
> >
> >What I propose is to sync the current status of QTextLayout/QTextEngine
> >from
> >Qt5, put it into a patchset for Qt4 and get it into the next Qt4 release.
> >
> >Would this be accepted? Or is there still a policy which forbids
> >performance
> >improvements to be merged into Qt4? Considering that the above makes the
> >Kate
> >editor practically useless for some files (freezes for multiple minutes),
> >I
> >personally see this as a bug fix anyways...
> >
> >Bye
--
Milian Wolff
mail at milianw.de
http://milianw.de
More information about the Development
mailing list