[sugar] Re: Problem building sugar-jhbuild

[Marco Pesenti Gritti]

On Sun, 2007-01-21 at 15:21 +1100, msevior at physics.unimelb.edu.au wrote:
> > On Fri, 2007-01-19 at 12:44 -0300, Gonzalo Delgado wrote:
> >>  While building pycairo:
> >>
> >> pycairo-context.c: In function 'pycairo_get_dash':
> >> pycairo-context.c:353: error: too many arguments to function
> >> 'cairo_get_dash_count'
> >> pycairo-context.c:358: error: void value not ignored as it ought to be
> >> pycairo-context.c: In function 'pycairo_get_dash_count':
> >> pycairo-context.c:386: error: too many arguments to function
> >> 'cairo_get_dash_count'
> >
> > It works for me. Have you built both cairo and pycairo?
> >
> > Otherwise the lib64 issue might be hitting you. Verify which cairo.h are
> > you actually using here, and if the signature for this function is the
> > expected one (cairo_get_dash_count (cairo_t *cr, int *count);) there.
> >
> 
> I get exactly the same error on an ubuntu edgy build for a standard 32 bit
> intel machine.
> 
> 
> The problem is that the API for cairo.h is different from what pycairo
> expects.
> 
> I changed:
> 
> cairo_get_dash_count (o->ctx, &count);
> 
> to:
> 
> count = cairo_get_dash_count (o->ctx);
> 
> In pycairo.
> 
> This is clearly a problem in pycairo and the sort of stupid bug we're all
> constantly fighting with. We abiword hackers are no better and have broken
> the sugar-jhbuild a few times.
> 

https://bugs.freedesktop.org/show_bug.cgi?id=9715

> Is there a way to set up some sort of frozen state of known working builds
> of each of the modules?

I'm working on a sugar-jhbuild (of our tiny layer over jhbuild) rewrite
which I hope will solve also this issue. Will be on git head in a couple
of days.

> 
> the alternative is to constantly shame the maintainers of the libraries to
> keep their builds unbroken.

I can understand builds breaking in unstable modules (like pycairo),
what really disappoint me is continuous breakage in the stable
branches :(

Marco