[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-commits
Subject:    Re: KDE/kdelibs/kdecore/sonnet
From:       Zack Rusin <zack () kde ! org>
Date:       2009-11-30 23:48:54
Message-ID: 200911301848.54364.zack () kde ! org
[Download RAW message or body]

On Monday 30 November 2009 18:38:18 David Faure wrote:
> On Tuesday 01 December 2009, Zack Rusin wrote:
> > On Monday 30 November 2009 17:53:12 David Faure wrote:
> > > On Monday 30 November 2009, Zack Rusin wrote:
> > > > SVN commit 1056377 by zack:
> > > >
> > > > sonnet: rewrite text segmentation algorithm
> > > >
> > > > since its creation sonnet was broken with indic, asian, arabic and
> > > > a lot of other languages that didn't use english-like alphabets.
> > > > this commit removes the custom text segmentation algorithm and
> > > > replaces it with a proper unicode tr29-11 algorithm found in
> > > > qtextboundaryfinder and hopefully makes sonnet work with all
> > > > languages in the world. let me know if causes any regressions.
> > >
> > > Well, that's what unittests are for ;-)
> >
> > Yea, it's why I added some when committing this unfortunately it's a bit
> >  hard to cover peculiarities of every language out there =)
> >
> > > It seems your changes introduce a regression, because
> > >  sonnet/tests/test_core uses 100% CPU for a very very long time and
> > > never terminates.
> >
> > Hmm, it works here. That test always took a bit of time. I just committed
> > a trivial fixlet that should improve it though, does it make it better
> > for you?
> 
> Yep, seems to work now.
> 
> BUT: this code was committed at the worst possible time (*very* close to
> beta1 tagging) and without review, breaking the freeze.
> And obviously (from your own words) it's impossible to be sure the code
> doesn't introduce any regressions. Can you revert it for now and commit it
> again a month from now when trunk is open for kde-4.5?
> I don't want to think about what will happen if we ship 4.4 with infinite
>  loops in some languages :/

To be honest I'm not sure if that makes much sense. If the code breaks 
something, then well it breaks something and if you don't have enough tests 
you most likely have to wait for a beta for it to be tested by general public. 
So yea, we can revert it and then let it sit and wait for kde-4.5 beta so that 
it's tested then or we can let it be tested with kde-4.4 beta. The code was 
broken for years, it just happened to work in enough cases for english, german 
and other latin languges which meant that no one cared enough about the other 
few billion people on the planet to fix it, at least now breakage in this 
breaks all languages equally.
Of course at the end of the day it's your call and if it will make you sleep 
better at night then sure revert it.

z
[prev in list] [next in list] [prev in thread] [next in thread]