gmane.comp.kde.devel.core

Re: qt-copy patch for QDesktopServices to allow path kcm to work

Oswald Buddenhagen — 2008-12-02T00:59:22

we already tried pushing it into the standard - to no avail so far. i think we'd have better chances to push a new $v which allows only variable expansion, unlike $e which also allows command substitution. and it should be obvious why we cannot use it in shared files until it is an accepted standard and all major players actually implement it.

Re: qt-copy patch for QDesktopServices to allow path kcm to work

Aaron J. Seigo — 2008-12-02T00:47:16

it's non-standard but ... it's a standard we created (ostensibly, anyways). but then we say "we can't do it because we have to stick within this standard" rather than try and improve that standard. seems a bit daft not to?

Re: qt-copy patch for QDesktopServices to allow path kcm to work

Oswald Buddenhagen — 2008-12-01T23:55:24

yes, but the actual implementation is non-trivial (at least the full one including $() syntax). apart from that, given that it is non-standard, kde definitely MUST NOT (yay, RFC terminology) write it to potentially shareable .desktop files. that was alread fixed (maybe even multiple times), but somehow it got re-enabled directly or indirectly, or this particular case was not covered in the first place.

Re: qt-copy patch for QDesktopServices to allow path kcm to work

Aaron J. Seigo — 2008-12-01T23:45:38

nothing else can read it because even when offered patches they reject them. *cough* what [$e] and [$i] provides is rather useful, and it would seem to make sense to either propose an alternative or adopt what is really an improvement.

KStyle/KIconEngine problem WAS Re: Bug in icon kcm?

Albert Astals Cid — 2008-12-01T23:31:24

A Dimarts 02 Desembre 2008, Albert Astals Cid va escriure: The problem is that it does not work though, most of the painting is done in QCommonStyle by calling KIconEngine::pixmap that always sets KIconLoader::Desktop as group because it seems it can not know which it's parent is. One "solution" would be never calling QCommonStyle and making KStyle deal with it all the calls so it can do the correct calls, but that's completely an impoosible amount of work. Other hack would be having a type variable in KIconEngine that is set each time KStyle calls QCommonStyle so when QCommonStyle ends up calling KIconEngine::pixmap we know which type of element we are rendering. But that would not work with threads, although i'm not sure if one can use KStyle in threads The obviously simple solution is dropping this icon effects thing altogether :-/ Albert

Re: Bug in icon kcm?

Albert Astals Cid — 2008-12-01T23:18:36

A Dilluns 01 Desembre 2008, Louai Al-Khanji va escriure: Done, thanks for reporting. Albert

Re: [ANNOUNCE] new website for kde mailing-list archives

David Faure — 2008-12-01T23:01:10

Done.

Re: KProcess Line Mode Patch

Thiago Macieira — 2008-12-01T20:53:39

It should have a "set" prefix. The getter can have an "is" prefix: void setLineModeEnabled(bool enable) bool isLineModeEnabled() const;

Re: qt-copy patch for QDesktopServices to allow path kcm to work

Dan Meltzer — 2008-12-01T20:53:13

On Mon, Dec 1, 2008 at 4:45 PM, Anne-Marie Mahfouf free.fr> wrote: I'm not the worlds greatest expert on regular expressions.. but why do you make the addition a group? wouldn't it be better to do ("^XDG_(.*)_DIR.*=(.*)$")) and that way it is not necessary to change the line further on.

Re: qt-copy patch for QDesktopServices to allow path kcm to work

Thiago Macieira — 2008-12-01T20:52:23

No, I don't think so. This patch will be rejected in Qt. KDE should be fixed to not add the [$e] because nothing other than KDE will be able to read it.

qt-copy patch for QDesktopServices to allow path kcm to work

Anne-Marie Mahfouf — 2008-12-01T21:45:44

Re: move QCoreApplication::addLibraryPath call from KApplication toKComponentData

David Faure — 2008-12-01T20:44:51

Looks good to me. Thanks.

Re: KProcess Line Mode Patch

Ingo Klöcker — 2008-12-01T20:41:41

Just a general advice that came to my mind while reading your message: The name of a method taking a bool parameter should clearly convey the meaning of passing true/false to the method. setOutputLineMode( bool mode ) fails to convey this meaning. This method should rather be named enableOutputLineMode( bool enable ) or be replaced by a more general setOutputMode( OutputMode mode ). Regards, Ingo

Re: Idea: SkipDirtyOnIdenticalValues WriteConfigFlag for KConfigBase/ KConfig

David Faure — 2008-12-01T20:20:17

On the contrary - if you change a toolbar to "big" and then to "normal" again, it looks normal like any other toolbar. If you change the system default 6 months later, it would be really unexpected that this toolbar doesn't behave like all others. Well, KMainWindow and KToolBar definitely do use revertToDefault(), so it would be nice if it could be unbroken......

Re: Race condition with data loss in ReadWritePart

David Faure — 2008-12-01T20:08:28

Yes, at least. There's a comment against that in the code already. ReadWritePart::~ReadWritePart() { // parent destructor will delete temp file // we can't call our own closeUrl() here, because // "cancel" wouldn't cancel anything. We have to assume // the app called closeUrl() before destroying us. } Done now. Better late than never... I don't understand this one. ReadOnlyPart::openUrl() already calls closeUrl which is virtual so it does call ReadWritePart::closeUrl() [or your reimplementation of it].

Re: Various problems/bugs with actions/toolbars in aKParts::MainWindow

David Faure — 2008-12-01T19:50:56

This works nowadays, after all the fixes by ereslibre and myself, right? This sounds like a misuse of kactioncollections? Not putting the actions into the collections, or into the wrong one, or something like that. The dialog works in other apps afaik. Still the case? Still the case?

Re: [PATCH] thread-safe KQueuedDialog episode 2

Matthias Kretz — 2008-12-01T19:01:22

Looks good to me. On Friday 28 November 2008 22:54:43 Sebastian Trueg wrote:

Re: KProcess Line Mode Patch

Thiago Macieira — 2008-12-01T18:14:28

We did. But I don't remember doing the work in QTextStream and QDataStream. So the solution is there, but not used.

Re: closing bugs products

Friedrich W. H. Kossebau — 2008-12-01T16:53:40

Hi, Hurra for all cleaning up of bko, in general :) Am Montag, 1. Dezember 2008, um 16:52 Uhr, schrieb Matt Rogers: Why shouldn't it stay a product of its own? Promo people still think of bringing it back to life, AFAIR. It is. Only part of KDE since 4.1. See e.g. here: http://utils.kde.org/projects/printer-applet/development.php There should perhaps be one central database for the metadata of all KDE programs/modules, which is reused in techbase, userbase, bugzilla and elsewhere, containing data like homepage, maintainer, releases, icon/logo, etc. So such information is consistent and can be automatically made available. Cheers Friedrich

Re: closing bugs products

Matt Rogers — 2008-12-01T15:52:14

isn't printer-applet new though?

Re: KProcess Line Mode Patch

Andreas Pakulat — 2008-12-01T15:12:03

Thiago and Ossi already commented on all the technical problems and wether this is really so much needed. Sure having some line-splitting in kdelibs might be good, but splitting a QByteArray at '\n' and returning the lines + "rest" is a simple function in a namespace (or working on QString instead) no need to fiddle around with delimiters, encodings and so on on a KProcess level. So -1 from me. Andreas