[PyKDE] PyKDE for sip 3.5 and 3.6
Phil Thompson
phil at river-bank.demon.co.uk
Fri Apr 18 23:49:00 BST 2003
On Friday 18 April 2003 9:20 pm, Jim Bublitz wrote:
> On 18-Apr-03 Gerard Vermeulen wrote:
> > On Fri, 18 Apr 2003 09:47:34 -0700 (PDT)
> >
> > Jim Bublitz <jbublitz at nwinternet.com> wrote:
> >> I don't believe there's an easy way to check which version of
> >> sip was used to build PyQt (please correct me if I'm wrong -
> >> I'd be happy to test for that).
> >
> > Require that the output of sip -V corresponds with the output of
> > python -c 'from qt import *; print PYQT_VERSION' (this still
> > works with 3.5). The problem is what to with snapshots: that is
> > the reason for 'used to check'.
>
> At the moment I'm not sure I ever did that - it should be done, as
> this has been a low level problem forever. That seems like a good
> place to issue a warning, as described below.
Be aware that the linkage between SIP and PyQt versions won't last much
longer. Fairly soon there will be one version of PyQt but two versions of SIP
- and it will be up to the user/packager to decide which SIP is to be used.
> > Well, I think that from a psychological point of view it is
> > better to have a proud user saying that they 'fixed' my build
> > tools and that PyQwt works with Python-5.0 than an unhappy
> > message 'it does not build with Python-5.0'. Anyhow, a warning
> > 'you are moving into unexplored territory: RTFM or you will die'
> > is better than g++ dying after 30 minutes of compilation
> > with an cryptic error message (some people have tried days before
> > writing to the list).
>
> I agree in principle, but I'm not optimistic about it working out
> that cleanly. I'm intimately familiar with the pain of compiles
> dying after 30+ minutes and don't wish that on anyone.
>
> >> > Another idea is to upgrade only PyQt's sip files, if a new Qt
> >> > release comes out (of course it is more work for Phil).
> >>
> >> I'm not sure what you mean here.
> >
> > Well, with 3.6 around the corner there is no immediate need,
> > but in the future it may be worthwhile to backport sip files
> > adapted to post-3.6-Qt-releases from post-3.6-snapshots to make
> > a 3.6.1 release.
> >
> > At least 50 % of recent mailing list traffic is related to
> > incompatibilities between PyKDE-3.5-2 and PyQt/sip snapshots
> > provoked by post-3.5-Qt-releases.
>
> I'll take credit for most of that - the combination of a long delay
> in getting the release out plus releasing prematurely anyway is
> most of the problem. What I did in the past that avoided this is
> delay the PyKDE release until the next sip release was ready -
> waiting didn't seem like a good choice this time, but in
> hindsight...
>
> (In my own defense - the incompatibilities in 3.6 didn't occur
> until right before or right after the release, but I got plenty of
> warning from Phil too)
>
> > Backporting from snapshots to PyQt-3.5 would have saved the world
> > time, but maybe not for Phil :-/
> >
> > And if Phil does not want to do it: a volunteer may want to earn
> > some fame :-)
>
> I think the better solution would be keep PyKDE in closer sync with
> PyQt (although that perhaps doesn't help PyQwt much) and match the
> PyQt snapshots with PyKDE snapshots within a reasonable period of
> time.
The current situation is very unusual. The change between Qt 3.1.1 and 3.1.2
is the first time that Trolltech have broken source level compatibility -
it's a bug. Coincidentally Python 2.3 also breaks source level compatibility,
and this is the first time that has happened.
I've also been (wrongly with hindsight) holding back on releasing 3.6 because
I see it as a great synchronisation point between up to date versions of
PyQt, Qt, Python, KDE, PyKDE, QScintilla and eric.
Anyway, in a couple of weeks or so, everything will have settled down and
everybody will have something solid to work with.
Phil
More information about the PyQt
mailing list