Generated changes and patch systems
On Thu, 2008-05-29 at 11:56 -0700, Russ Allbery wrote:
> Simon McVittie <smcv@debian.org> writes:
>
> > Here's how gtk-doc *used to* work:
> >
> > * gtk-doc parses source code and writes out skeletal tmpl/*.sgml
> > * svn ci -m 'initial version of gtkdoc templates' tmpl
> > * upstream doc author inserts content into tmpl/*.sgml
> > * svn ci -m 'wrote some docs' tmpl
> > * gtk-doc parses source code (to pick up any new functions) + tmpl/*.sgml,
> > and merges them
> > * svn ci -m 'yay gtkdoc' tmpl
> > * ...
> >
> > This is, as you've noticed, insane. Sane upstreams now write all the
> > content for the documentation in /** */ comments in the source code, so
> > the tmpl/*.sgml are purely generated and can safely be omitted from
> > source-code control. (I have no opinion on whether your upstream is sane
> > or not - please check.)
> >
> > However, the "no rule to make tmpl/*.sgml" issue still exists, as a
> > relic of the old build process.
>
> Thank you for the explanation! This now makes much more sense.
Same here - I'm glad there was an explanation for this behaviour, it was
driving me nuts.
;-)
> Sounds to me like the first thing to try would be to just regenerate all
> of the tmpl/*.sgml files via gtk-doc and delete them in the clean rule and
> see if that works properly for this project. That's what I'd do, at
> least; that ensures a clean build without putting artifacts in the
> *.diff.gz.
Got a few other things to do first but yes, I will explore that before
uploading the new upstream version. Thanks both.
--
Neil Williams <codehelp@debian.org>
|
