On Fri, 2025-10-24 at 18:01 +0200, sj wrote:
> I couldn't find the time to look into Autotools before today, but I 
> think I've found a working solution. Let me know if there's something
> that still needs fixing, but this works on my machine.
> 
> 0001-*.patch contains the --enable-atril configure option, which is
> now 
> the default and it looks for atril direcly, while --enable-evince
> only 
> looks for evince and not atril. I think this is a bit more intuitive
> and 
> needs a lot less configure.ac reading to figure out. If both are 
> enabled, it's as if only --enable-atril was given (to avoid making a 
> decision on test order).
> 
> 0002-*.patch builds on the first one (they probably commute but I 
> haven't tried, it should an easy fix anyway if they don't) and logs a
> deprecation warning every time GTK2 is used. I also disabled 
> autodiscovery, so now Denemo doesn't compile if GTK3 is unavailable, 
> unless GTK2 is explicitly enabled, which seems like a fair middle
> ground 
> between breaking support in a single commit and not doing anything.
> With 
> my changes people still stuck on GTK2 builds for no reason other than
> missing headers need to at least add a single build flag or install
> the 
> GTK3 development libraries.
> 
> I tested with GTK3 and atril, also GTK3+evince, which gives me the
> same 
> error about PDFs as before, so it works I guess but someone who has a
> working evince version should probably check.

Wonderful! These are both in the repository now.

Richard

> 
> sj
> 
> On 10/4/25 13:42, sj wrote:
> > I agree about GTK2 (not to mention that there's probably no one to
> > test 
> > it anymore, which means additional maintenance overhead for no
> > reason), 
> > also the atril switch. I'm no autotools wizard either, but I can
> > try 
> > later tomorrow.
> > 
> > On 10/4/25 11:27, Richard Shann wrote:
> > > On Thu, 2025-10-02 at 20:31 +0200, sj wrote:
> > > > I think --enable-atril isn't even a configure option, USE_ATRIL
> > > > is
> > > > only defined if --enable-evince is true but the evince
> > > > development
> > > > packages are missing so it falls back to Atril. Actually,
> > > > USE_ATRIL
> > > > is defined in an if test "x$have_evince" != xno block (this is
> > > > just
> > > > the configure option), if a use_atril variable is also on,
> > > > which is
> > > > only set if the evince checks failed but it finds atril,
> > > 
> > > I think Denemo should default to atril and only use evince if it
> > > is not
> > > available, as evince has failed to honor the point and click in
> > > PDF for
> > > years now, which limits its usefulness in Denemo. Unfortunately
> > > I've
> > > never managed to get my head around autotools so I'm unsure how
> > > to
> > > arrange this.
> > > 
> > > >   at least in the GTK3 branch. I think this is the same for
> > > > GTK2 but I
> > > > haven't looked at it too much.
> > > 
> > > I think GTK2 is now old enough to be forgotten...
> > > 
> > > Richard
> > > 
> > > > 
> > > > 
> > > > Ekkor: 2025. október 2. 19:13:49 CEST, Andreas Schneider
> > > > <[email protected]> írta:
> > > > > Am 02.10.25 um 13:02 schrieb Richard Shann:
> > > > > > On Wed, 2025-10-01 at 19:49 +0200, Andreas Schneider wrote:
> > > > > > > Am 01.10.25 um 14:05 schrieb Richard Shann:
> > > > > > > > On Tue, 2025-09-30 at 20:26 +0200, Andreas Schneider
> > > > > > > > wrote:
> > > > > > > > > Richard, can you please apply the patch from
> > > > > > > > > http://deb.debian.org/debian/pool/main/d/denemo/ 
> > > > > > > > > denemo_2.6.39-03.debian.tar.xz
> > > > > > > > >     to make Denemo compile with recent libevince? You
> > > > > > > > > can
> > > > > > > > > also find
> > > > > > > > > it
> > > > > > > > 
> > > > > > > > Hmm, I get an error trying to build with this,
> > > > > > > > 
> > > > > > > > ../src/source/proof.c:283: undefined reference to
> > > > > > > > `EV_DOCUMENT_ANNOTATIONS'
> > > > > > > > 
> > > > > > > > Is this something not in the atril headers or is it a
> > > > > > > > version
> > > > > > > > matter?
> > > > > > > 
> > > > > > > Have you installed the libatrildocument-dev package?
> > > > > > 
> > > > > > well,
> > > > > > dpkg-query -l yields
> > > > > > 
> > > > > > ii  libatrildocument-dev                         1.26.0-
> > > > > > 2+deb12u3                     amd64        MATE document
> > > > > > rendering library (development files)
> > > > > > 
> > > > > > and I see that the inclusion of
> > > > > > libdocument/ev-document-annotations.h
> > > > > > is missing in atril-document.h so this seems to be a bug in
> > > > > > libatrildocument.dev
> > > > > > 
> > > > > > To work around this bug I propose the attached patch. I'll
> > > > > > apply
> > > > > > this
> > > > > > unless there are objections.
> > > > > 
> > > > > Thank you very much, the patch looks good, and it enables me
> > > > > to
> > > > > compile successfully with libevince.
> > > > > 
> > > > > When I try to compile with --disable-evince --enable-atril, I
> > > > > get
> > > > > the error
> > > > > 
> > > > > pageturner.c:28:10: fatal error: evince-view.h: Datei oder
> > > > > Verzeichnis nicht gefunden
> > > > >    28 | #include <evince-view.h>
> > > > >       |          ^~~~~~~~~~~~~~~
> > > > > compilation terminated.
> > > > > 
> > > > > It seems the constant USE_ATRIL is not defined correctly.
> > > > > What have
> > > > > I made wrong?
> > > > > 
> > > > > Andreas
> > > > > 
> > > > 
> > > 
> > > 
> > 
> > 


Reply via email to