Re: new glib2 c++ template errors in inkscape,gegl,wireshark

[Patrick Welche <prlw1%cam.ac.uk@localhost>]

On Wed, Mar 24, 2021 at 09:09:43AM +0100, Thomas Klausner wrote:
> On Tue, Mar 23, 2021 at 03:28:12PM +0000, Patrick Welche wrote:
> > On Tue, Mar 23, 2021 at 11:39:18AM +0100, Thomas Klausner wrote:
> > > On Tue, Mar 23, 2021 at 10:29:17AM +0000, Jonathan Perkin wrote:
> > > > * On 2021-03-23 at 10:04 GMT, Martin Husemann wrote:
> > > > 
> > > > > Pretending it is a simple update, IMHO, is not a very efficient way to
> > > > > deal with it.
> > > > 
> > > > It also completely broke the build on SunOS.  I don't really understand how
> > > > this wasn't considered a low-level library, as per freeze rules?
> > > 
> > > It was committed before the freeze.
> > > 
> > > I've asked adam to back this out for now.
> > 
> > Thank you to adam@ for backing out his glib2 update.
> > 
> > The gobject-introspection update should also be backed out - it
> > contains .gir files for glib2, so will contain symbols which don't
> > exist yet.
> 
> Can this cause breakage, and if so how?

The theory goes as follows: programs which use Gnome libraries, but which
use a language other than C, discover the symbols of those of libraries
in the libraries' .gir files, which are created by gobject-introspection
when the libraries are built. This is fine, except in the case of glib, 
which gobject-introspection depends on. Rather than create a circular
dependency between glib and gobject-introspection (which I think existed
for a time(!)), gobject-introspection ships the .gir file for glib.
This means that those 2 packages should be in synch, otherwise the
.gir file in gobject-introspection could claim the existence of symbols
which don't actually exist in the glib library.


Cheers,

Patrick