version.h is not regenerated

[Dirk Hohndel]

On Wed, 2013-10-16 at 21:36 +0200, Cristian Ionescu-Idbohrn wrote:
> On Wed, 16 Oct 2013, Henrik Brautaset Aronsen wrote:
> > Cristian Ionescu-Idbohrn wrote:
> > > On Wed, 16 Oct 2013, Cristian Ionescu-Idbohrn wrote:
> > > > $ grep '^version\.h' .gitignore
> > > > $ rm -f version.h
> > > > $ git diff
> > > > $ make
> > > > make -C .../Documentation
> > > > OUT=.../Documentation/ doc
> > > > make[1]: Entering directory
> > > > `.../Documentation'
> > > > make[1]: Nothing to be done for `doc'.
> > > > make[1]: Leaving directory
> > > > `.../Documentation'
> > >
> > > Does ssrf-version.h replace version.h?
> >
> > Yes.
> 
> What was so wrong with version.h?
> After all, many packages use such thing as version.h.

The brilliant dependency generation in qmake wasn't able to tell apart
"version.h" and "libdivecomputer/version.h". So anything that depended
on the divecomputer code was rebuilt whenever I committed a change.

Granted, for most people in most situations this really didn't matter...
for me it was really annoying when I test commits. So I decided to fix
this the easy way

/D