[Kde-bindings] Re: PyKDE 4.5.2

Hans-Peter Jansen hpj at urpla.net
Fri Nov 26 17:46:17 UTC 2010


On Friday 26 November 2010, 17:28:03 Arno Rehn wrote:
> On Friday 26 November 2010 17:17:39 Hans-Peter Jansen wrote:
> > Hi,
> >
> > current PyKDE from kdebindings-4.5.2 doesn't build anymore.
> >
> > First, I needed to relocate the akonadi Entity dtor to the
> > protected section (see attached patch), but then it fails due to
> > some smokegen error resulting in:
> > [...]
> > What's the rationale of replacing sip here? Is it really possible
> > to mix python bindings of different generators with no apparent
> > downsides? How are Qt properties, signals/slots, QStrings,
> > QVariants and the like handled, when they cross the generators
> > borders? (IOW: interoperability) Is there any smokegen libraries
> > reference guide available (similar to pyqt4ref.html)?
>
> Smoke doesn't have anything to do with PyKDE or Python bindings. It's
> only needed for Ruby, C#, Perl and some other languages.
> I guess you're building against trunk? (If you're not, then the API
> of Nepomuk has changed in a minor release... that should not happen,
> normally). Building an old version of the bindings against a new
> version of the API should generally not considered to be supported.
> 'Old' bindings binaries however should continue to work. Just
> _building_ against new versions is not supported.
>
> Grab kdebindings from trunk, they already have the SmokeNepomuk fix.
> Unfortunately I can't help you with the SIP error since I don't do
> Python at all.

I found it: since some smoke bindings were created in my PyKDE build, I 
was under the impression, that parts of it are generated by smokegen 
now, and parts are sip based (which would be a disaster software 
architecture wise). Sorry about that.

Apart from the akonadi fix, I've just needed some more spec tweaks to 
suppress the smoke generator build..

Pete



More information about the Kde-bindings mailing list