Public SourceMage Discussion List

[Remko van der Vossen <wich AT yuugen.jp>]

On Wed, Aug 24, 2011 at 10:52:21AM +0200, Bor Kraljič wrote:
> On 24. MES 2011 10:42:49 Remko van der Vossen wrote:
> > On Wed, Aug 24, 2011 at 10:33:56AM +0200, Jaka Kranjc wrote:
> > > On Wednesday 24 of August 2011 10:20:42 Remko van der Vossen wrote:
> > > > On Tue, Aug 23, 2011 at 11:40:02PM +0200, Jaka Kranjc wrote:
> > > > > We should also rename the old kde spells to kde3 and the kde4 to 
> > > > > plain
> > > > > kde (later) — upstream is already dreaming about kde5, so it is 
> > > > > best to
> > > > > lose the version suffix for the latest branch or we'll rerun into 
> > > > > this
> > > > > issue each major release.
> > > > 
> > > > First off, let me say that I don't use KDE, or any DE for that matter 
> > > > as
> > > > most of you will know. However, I'd still like to add my 2 cents.
> > > > 
> > > > Is it wise to rename all kde4 spells to kde? We'll only get the same
> > > > problems upon the next KDE release. When it comes to things like KDE 
> > > > and
> > > > gnome it's more like every new major version is a completely new
> > > > project, so why not treat it as such. We could have the kde spell 
> > > > simply
> > > > be a profile spell that lets the user choose between any versions we
> > > > have available currently, like we do for instance with the ruby 
> > > > profile
> > > > spell and the ruby-1.8 and ruby-1.9 spells.
> > > Ruby is just one spell, there'd need to be a dozen profiles for kde. 
> > 
> > I don't think so, it'd only need to be done for the main kde profile
> > spells, once it's installed people can figure out they need kde#-$app,
> > we don't treat our users as idiots as far as I am aware...
> For me it came to simple question. WILL WE SUPPORT KDE3? (and if yes, WHO 
> will support it. It won't be 
> me.) It isn't supported by upstream. qt3 (qt-x11) that is needed for kde3 
> isn't supported by upstream. 
> So why should we support kde3? ATM we don't even have the information if 
> kde3 compiles with 
> current version of gcc. 
> 
> If we decide not to support kde3 and remove it from grimoire we avoid 
> problems with renaming. Since 
> after a while it would be quite easy to rename kde4 to kde and continue 
> without suffix. But somehow I 
> don't think we will agree on removal of kde3. Since from unknown reason we 
> REALLY like having VERY 
> old spells in grimoire for which we don't even know if they compile. 

I'm not at all arguing for or against removal of KDE3, I'm saying,
whatever happens, I think it'd be better to just keep kde4-* named
kde4-* and not rename the whole shebang to kde-*, doing so may only
serve to cause us trouble.

One such scenario, if we decide to remove KDE3, we should not deprecate
the spells, but just outright remove them from grimoire and inform the
users. That way people who have it installed will not lose it, but will
be informed that we no longer support it and there will never be an
upgrade, hopefully urging them to upgrade to KDE4, or perhaps something
even better. ;)

However, were we to later rename the kde4-* spells to kde-* we'd destroy
this scenario as people would have installed KDE3 software magically and
incompatibly upgraded to KDE4 software because the spell names were the
same.

The other way, i.e. keeping KDE3 and renaming the kde-* spells to
kde3-*, then no delay between the kde -> kde3 and kde4 -> kde renames is
going to be enough, there will always be people who will skip a number
of stables for whatever reason, perhaps simply because they installed
from iso, cast KDE from unupdated grimoire and then update to latest
stable grim. Stuff'd break.

To guard against these kinds of things we should add code to each and
every kde4-* -> kde-* renamed spell to check whether the spell is
already installed and if so, if it's a version from KDE3, abort the
cast. In other words, a crap load of work to do it right.

If on the other hand we simply keey kde4-* be kde4-* we have no such
issues, and we will never have such issues in the future if we do the
same with 5, 6, 7, etc.

Sorry for the rant,

Remko.

Attachment: pgpx54rTqHZet.pgp
Description: PGP signature