Despite the temperature being 35-40C all of last week - it cools
down nicely at night, but my office was quite warm by late
afternoon, and my computer doesn't work when the CPU temperature
hits 81C - and an 8 hour power outage last night[1], PyKDE for
KDE312/sip3.6/sip3.7 is coming along. KDE311 builds against 3.6
and 3.7 with no changes; I'm just finishing the final test build
for KDE312/sip 3.6 and expect no problems going to sip 3.7. I
want to test building against earlier KDE versions and then
package it, so I hope to have it available by Wednesday, and
certainly no later than the end of the week.
Requirements
==========
sip 3.6 **or** sip 3.7
PyQt >= 3.0
KDE >= 3.0
1.5.2 <= Python <= 2.2.2 (2.3 is untested - should work)
Note that I've now dropped support for KDE2 (I indicated this
would happen about six months ago) and Phil will be dropping Qt2
support in the future, so I'd just as soon get this over with
now - it saves a couple days of build testing as the computers
still running KDE2 are old and slow (but then so am I).
I've only tested on SuSE 8.2 (gcc 3.3) so far and haven't had any
problems. I'll hit some earlier gcc versions with the earlier
KDE3 versions. I have RH9.1 and Mdk 9.1, but not installed -
I'll postpone testing against those until after this release.
I haven't installed Python 2.3 yet, but will get around to that
after the release. I don't expect problems there, unless sip or
PyQt are having problems as well.
I've tested KDE311 and 312 against sip 3.6 and they build fine.
All other tests will be against sip 3.7.
Bugs and Patches
=============
I've only patched PyKDE far enough to work with sip 3.6 and sip
3.7. There are a bunch of bug reports and patches pending from
the list (Pete's SMP patch, a bunch of stuff Gordon Tyler worked
on, and some other stuff). I'll be getting to that next (after
the release). I expect to do a release every week or two for a
while, so packagers might want to be aware of that. The
numbering will be of the form PyKDE-3.7-X, where X = 1, 2, 3 ...
The example programs work; there are unit tests for most of the
handwritten code - 3 fail (worked previously) One fails because
the method it's testing is obsolete and I haven't updated yet.
The other two failures are on methods not likely to be used by
anybody, so I'll put those off a while too. For one, Python
can't find the method ("missing attribute") and for the other,
the return value is empty, despite identical code on similar
methods still working fine. I'll try to remember to identify
these in BUGS - I did verify that they're actual failures (not
incorrect tests).
UPDATE: The KDE311 'return value' failure works with KDE312.
Don't ask me why - it's the same code. Both might be KDE or SuSE
problems.
Other
====
There are a few other minor rough spots but they have workarounds
in place. The docs are essentially unchanged from KDE311, so I
probably won't update those for the initial release.
build.py will no longer require the '-l' switch (KDE3 requires
libqt-mt) but will include a test for libqt-mt. I'll also
probably make the -c switch the default and require a '-c-' to
disable it (-c concatenates all of the source files for each
module into a single huge cpp file that compiles about 80%
faster, but requires 128MB or more of RAM to be useful). The
current basic command line:
python build.py -lqt-mt -c
will still work. -lqt however will cause an error, and only -c-
will have an effect, and 'python build.py' will work the same as
the command line above.
As always, check the README and ChangeLog.
Jim
[1] My servers had been up 459 days 12 hours when I had to shut
them down. Ugh.
_______________________________________________
PyKDE mailing list [EMAIL PROTECTED]
http://mats.imk.fraunhofer.de/mailman/listinfo/pykde
