Libguestfs - Dec 2013 - Re: [PATCH 2/3] po: fix dependencies for libguestfs.pot extraction

On Thu, Dec 05, 2013 at 04:30:05PM +0100, Pino Toscano
wrote:
> Fix the dependencies of the libguestfs.pot target: other than using the
> right make variables holding the contents of the POTFILES, depend also
> on the POTFILES themselves.
> ---
>  po/Makefile.am | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/po/Makefile.am b/po/Makefile.am
> index b0a8038..a8343ec 100644
> --- a/po/Makefile.am
> +++ b/po/Makefile.am
> @@ -65,7 +65,7 @@ XGETTEXT_ARGS = \
>  	--msgid-bugs-address="$(MSGID_BUGS_ADDRESS)" \
>  	--directory=$(top_srcdir)
>  
> -$(DOMAIN).pot: Makefile $(POTFILES) $(POTFILES-pl) $(POTFILES-ml)
> +$(DOMAIN).pot: Makefile POTFILES $(POTFILES) POTFILES-pl $(POTFILES_PL)
POTFILES-ml $(POTFILES_ML)
>  	rm -f $@-t
>  if HAVE_OCAML_GETTEXT
>  	$(OCAML_GETTEXT) --action extract --extract-pot $@-t $(POTFILES_ML)
So I agree that $(POTFILES-pl) is definitely wrong.  Not sure exactly
what we were thinking about there ...

But, won't the addition of the literal file names break separate
compilation?  In particular, $(POTFILES_PL) is supposed to be the
correct path to the file POTFILES-pl (and correspondingly for the
other files), so it shouldn't be necessary to list both POTFILES-pl
and $(POTFILES_PL).

Separate compilation is something that Debian cares about a lot
because they have to rebuild libguestfs several times in series for
different python versions etc.

Rich.

-- 
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
Fedora Windows cross-compiler. Compile Windows programs, test, and
build Windows installers. Over 100 libraries supported.
http://fedoraproject.org/wiki/MinGW

On Thursday 05 December 2013 15:53:58 Richard W.M. Jones
wrote:
> On Thu, Dec 05, 2013 at 04:30:05PM +0100, Pino Toscano wrote:
> > Fix the dependencies of the libguestfs.pot target: other than using
> > the right make variables holding the contents of the POTFILES,
> > depend also on the POTFILES themselves.
> > ---
> > 
> >  po/Makefile.am | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/po/Makefile.am b/po/Makefile.am
> > index b0a8038..a8343ec 100644
> > --- a/po/Makefile.am
> > +++ b/po/Makefile.am
> > @@ -65,7 +65,7 @@ XGETTEXT_ARGS = \
> > 
> >  	--msgid-bugs-address="$(MSGID_BUGS_ADDRESS)" \
> >  	--directory=$(top_srcdir)
> > 
> > -$(DOMAIN).pot: Makefile $(POTFILES) $(POTFILES-pl) $(POTFILES-ml)
> > +$(DOMAIN).pot: Makefile POTFILES $(POTFILES) POTFILES-pl
> > $(POTFILES_PL) POTFILES-ml $(POTFILES_ML)> 
> >  	rm -f $@-t
> >  
> >  if HAVE_OCAML_GETTEXT
> >  
> >  	$(OCAML_GETTEXT) --action extract --extract-pot $@-t
> >  	$(POTFILES_ML)
> 
> So I agree that $(POTFILES-pl) is definitely wrong.  Not sure exactly
> what we were thinking about there ...
> 
> But, won't the addition of the literal file names break separate
> compilation?  In particular, $(POTFILES_PL) is supposed to be the
> correct path to the file POTFILES-pl (and correspondingly for the
> other files), so it shouldn't be necessary to list both POTFILES-pl
> and $(POTFILES_PL).
The idea behind the addition is make the libguestfs.pot generation 
dependent on both the POTFILES files and the actual sources, so either 
adding a new source to any POTFILES or changing any of the listed 
sources in any POTFILES will trigger a new pot rebuild.
> Separate compilation is something that Debian cares about a lot
> because they have to rebuild libguestfs several times in series for
> different python versions etc.
I guess for "separate compilation" you mean an out-of-tree, or 
generically when srcdir!=builddir? If so you are right, there was a bug 
in my patch as it didn't reference the POTFILES files with their path to 
the source directory. I just fixed it, new patch coming shortly.

-- 
Pino Toscano

Fix the dependencies of the libguestfs.pot target: other than using the
right make variables holding the contents of the POTFILES, depend also
on the POTFILES themselves.
---
 po/Makefile.am | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/po/Makefile.am b/po/Makefile.am
index b0a8038..a695adc 100644
--- a/po/Makefile.am
+++ b/po/Makefile.am
@@ -65,7 +65,7 @@ XGETTEXT_ARGS = \
 	--msgid-bugs-address="$(MSGID_BUGS_ADDRESS)" \
 	--directory=$(top_srcdir)
 
-$(DOMAIN).pot: Makefile $(POTFILES) $(POTFILES-pl) $(POTFILES-ml)
+$(DOMAIN).pot: Makefile $(srcdir)/POTFILES $(POTFILES) $(srcdir)/POTFILES-pl
$(POTFILES_PL) $(srcdir)/POTFILES-ml $(POTFILES_ML)
 	rm -f $@-t
 if HAVE_OCAML_GETTEXT
 	$(OCAML_GETTEXT) --action extract --extract-pot $@-t $(POTFILES_ML)
-- 
1.8.3.1

On Thu, Dec 05, 2013 at 07:02:28PM +0100, Pino Toscano
wrote:
> On Thursday 05 December 2013 15:53:58 Richard W.M. Jones wrote:
> > On Thu, Dec 05, 2013 at 04:30:05PM +0100, Pino Toscano wrote:
> > > Fix the dependencies of the libguestfs.pot target: other than
using
> > > the right make variables holding the contents of the POTFILES,
> > > depend also on the POTFILES themselves.
> > > ---
> > > 
> > >  po/Makefile.am | 2 +-
> > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > > 
> > > diff --git a/po/Makefile.am b/po/Makefile.am
> > > index b0a8038..a8343ec 100644
> > > --- a/po/Makefile.am
> > > +++ b/po/Makefile.am
> > > @@ -65,7 +65,7 @@ XGETTEXT_ARGS = \
> > > 
> > >  	--msgid-bugs-address="$(MSGID_BUGS_ADDRESS)" \
> > >  	--directory=$(top_srcdir)
> > > 
> > > -$(DOMAIN).pot: Makefile $(POTFILES) $(POTFILES-pl)
$(POTFILES-ml)
> > > +$(DOMAIN).pot: Makefile POTFILES $(POTFILES) POTFILES-pl
> > > $(POTFILES_PL) POTFILES-ml $(POTFILES_ML)> 
> > >  	rm -f $@-t
> > >  
> > >  if HAVE_OCAML_GETTEXT
> > >  
> > >  	$(OCAML_GETTEXT) --action extract --extract-pot $@-t
> > >  	$(POTFILES_ML)
> > 
> > So I agree that $(POTFILES-pl) is definitely wrong.  Not sure exactly
> > what we were thinking about there ...
> > 
> > But, won't the addition of the literal file names break separate
> > compilation?  In particular, $(POTFILES_PL) is supposed to be the
> > correct path to the file POTFILES-pl (and correspondingly for the
> > other files), so it shouldn't be necessary to list both
POTFILES-pl
> > and $(POTFILES_PL).
This ^^ is bogus.  I misunderstood the $(SED) in the macros.
Obviously $(POTFILES) expands to the contents of the POTFILES file,
ie. a list of filenames.
> The idea behind the addition is make the libguestfs.pot generation 
> dependent on both the POTFILES files and the actual sources, so either 
> adding a new source to any POTFILES or changing any of the listed 
> sources in any POTFILES will trigger a new pot rebuild.
Right, so I believe the first version (posted above) is correct,
because $(srcdir)/.. is not necessary (because automake is supposed to
add a VPATH for these builds).

Which apparently are called "VPATH builds", not separate compilation.

If you agree, I will push the first version of the 2/3 patch.

Rich.

-- 
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
virt-df lists disk usage of guests without needing to install any
software inside the virtual machine.  Supports Linux and Windows.
http://people.redhat.com/~rjones/virt-df/