Re: gobject-introspection build failure

[Colin Helliwell <colin helliwell ln-systems com>]

> On 03 April 2017 at 10:55 Thomas Haller <thaller redhat com> wrote:
>
> On Mon, 2017-04-03 at 10:00 +0100, Colin Helliwell wrote:
>
> > > On 30 March 2017 at 18:14 Colin Helliwell <colin.helliwell@ln-syste ms.com> wrote:
> > >
> > > > On 30 March 2017 at 17:34 Thomas Haller <thaller redhat com>
> > > > wrote:
> > > >
> > > > Hi,
> > > >
> > > > Not surprisingly, when cross compiling, you also need python and
> > > > pygobject in the other architecture.
> > > >
> > > > That makes it rather cumbersome.
> > > >
> > > > As I wrote in the other email, you can pre-generate those files
> > > > on
> > > > another host, during make dist and then build from source-
> > > > tarball.
> > >
> > > Thanks Thomas.
> >
> > Just to clarify Thomas: I can pre-generate on any other host? i.e.
> > even different architecture...?
>
> yes.

Thanks

> Optimally, you choose the same configure options (beside --enable-gtk-
> doc), because the man pages list default values that depend on them.

Noted - not tooooo critical a drawback at the moment.

So as to avoid the hazards of taking too many steps in one go, I'm thinking to first 'move up a version' from 
1.4.4 by using the 1.6.2 tarball. 
Just to explain my general aims: I'd ideally like to be even more up-to-date (e.g. tracking 1.8 master) but, 
given the cross-compile issues I have, maybe I need to defer that goal until there's a 1.8.x tarball
So if I can get going with 1.6.2 tarball then I'm a) slightly more up-to-date [I'll just have to watch out 
for patches that I need - not ideal, but ok]; and b) closer to being ready to jump on a 1.8.x tarball when it 
appears.
Does that sound reasonably sensible..?