[PyKDE] PyKDE 3.11 snapshot

[Jim Bublitz]

On Friday April 23 2004 04:00, Torsten Marek wrote:
> Jim Bublitz schrieb:

> > The deprecation warnings are coming from KDE itself (from
> > the h files) I would think.  I haven't seen them on my
> > compiles though. Right now I have that gcc feature
> > effectively #defined out when I generate the sip files from
> > KDE h files. If the methods were removed, I would be
> > versioning them out, but since they're still valid (just
> > deprecated) it seems to me I should leave them in.  I'm not
> > sure if supressing the warnings would be the right thing to
> > do either, but I hate compiles that generate warnings.

> > I also don't get the type errors you have (using 3.10.1) -
> > that may be due to differences in how sip 3.10.1 and sip 4.0
> > generate or use code. Most of the newer handwritten code
> > (which is where the errors are occurring) should have the
> > typecasts, but you can just add them:

> > sip/dcop/dcopobject.sip:136
> >         cpp = (DCOPObject *)sipForceConvertTo_DCOPObject
> > (elem, &iserr);

> > sip/dcop/dcopclient.sip:215
> >         cpp = (QCString *)sipForceConvertTo_QCString (elem,
> >                                                 &iserr);

> > and similarly for any other occurances. A quick grep turns
> > up 38 files that *might* need casts added, basically on
> > sipForceConvertTo_* calls. The cast will always be the same
> > as the class name after the '_' in the function name, and
> > always a pointer. Actually, it wouldn't be hard to write a
> > Python script to fix it. There may be some other places
> > where similar problems come up.

> > I'll be getting to sip 4.0 in a week or two, and I expect
> > most of the changes needed to be similar to what you've
> > found, so the conversion shouldn't take long. Most of the
> > changes for 3.10.1 are 4.0 compatible, although I've
> > probably missed a few. I'd like to finish the 3.10.1 version
> > before getting into 4.0 stuff.

> Since I got some spare time on my hands left (hopefully), I'll
> see if it's possible to fix this. If it is, I'll send you
> patches resp. the script for patching.

That'd be great. I'll be happy to provide any help I can. 

I think the only changes should be in handwritten code (plus the 
%Import stuff), and that's all C/C++.

I should see if there's a way to turn off the deprecation 
warnings, and then possibly add a switch to configure.py to 
disable them by default but allow them to be enabled. That seems 
like a reasonable solution to the problem if it can be done.

Jim