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

List:       kde-core-devel
Subject:    Re: Review Request: Ensure Plasma Desktop does not crash on exit
From:       Thomas =?UTF-8?B?TMO8Ymtpbmc=?= <thomas.luebking () web ! de>
Date:       2011-11-22 12:27:53
Message-ID: 20111122132753.0e0cca2f () web ! de
[Download RAW message or body]

Am Tue, 22 Nov 2011 11:49:39 +0100
schrieb Thiago Macieira <thiago@kde.org>:

> On Tuesday, 22 de November de 2011 09.57.32, Thomas Lübking wrote:
> > is there any detailed (official) explanation on this?
> 
> Yes.
> 
> In Qt, we declared code that accessed the list in question *while*
> the items are being deleted as broken by design, so we felt free to
> optimise qDeleteAll by ignoring people who did such evil things.

While I completely agree on the design aspect (and was/am no big
qDeleteAll user myself anyway), I -forgive my imprecision- actually
meant like "here":

http://doc.qt.nokia.com/4.8-snapshot/qtalgorithms.html#qDeleteAll

Not even "gg:qdeleteall site:blog.qt.nokia.com" showed anything (what
doesn't mean I count blogs as proper communication channel)

Given a simple mutex and a qWarning() -if hit- would not only have told
people but also prevented crashes by implementation rather than by
convention, I feel required to mention that this change of
implications -as done- seems a rather bold move which certainly caused
some stir and -at least on my side- worries about what actually *is*
the 4.8 qDeleteAll() problem (the reason why I clicked at this RR in
the first place)

Therefore a tiny bit red color would have been much appreciated.

Cheers,
Thomas

PS: to make that clear - I'm well aware that I moan on exceptionally
high standards here ;-)
[prev in list] [next in list] [prev in thread] [next in thread]