Re: texlive & evince
On Sat, Feb 09, 2019 at 07:54:37PM +, Edd Barrett wrote: > Hi, > > On Thu, Feb 07, 2019 at 05:34:35PM +0100, Jeremie Courreges-Anglas wrote: > > > Index: pkg/PLIST-synctex > > > === > > > RCS file: pkg/PLIST-synctex > > > diff -N pkg/PLIST-synctex > > > --- /dev/null 1 Jan 1970 00:00:00 - > > > +++ pkg/PLIST-synctex 4 Feb 2019 17:39:55 - > > > @@ -0,0 +1,15 @@ > > > +@comment $OpenBSD$ > > > +@conflict teTeX_base-* > > > +@conflict pdfjam-* > > > +@conflict ht-<2 > > > +@conflict texlive_texmf-buildset-<2018 > > > +@conflict texlive_base-<2018p1 > > > > The last line looks fine but I doubt that you need to copy the other > > @conflict lines. Were the -synctex files once provided by one of those > > packages? > > I looked at what espie@ did with the mktexlsr subpackage. > > The only conflict that might have conceivably included synctex was > teTeX_base, but the synctex source code has copyright starting 2008, and > teTeX died in 2006. So, I guess not. > > > As already discussed, I don't understand how it helps for those depends > > to be that strict. Aren't files more commonly moved between > > texlive/texmf subpackages, rather than from texlive/base to > > texlive/texmf? Unless I'm missing something, we may easily avoid the > > problem by only stating minimal versions: less chances for errors and less > > churn for pkg_add -u. Please find a diff below. > > Yep, sounds good. Diff also looks good. > > So are we saying that, with that change, the diff I posted to move > synctex into it's own sub-package would then be good to go (once we fix > the merge conflicts that will be caused by your change)? > > > Well, as mentioned by kili there's also the option to force evince to > > always use its internal copy. This means one more copy of synctex to > > watch for security/portability issues, but on the other hand this would > > decouple evince and texlive updates. Not a bad idea given the recent, > > undocumented API/ABI break in libsynctex... > > > > https://github.com/jlaurens/synctex/issues/23 > > My gut feeling says we should use the one from TeX live, but I think we > should hear from jasper@ and/or aja@, since they are the ones who > maintain evince. > > Guys, what would you prefer? I agree, we should use the one from TeX live. That's also what major Linux distros do as well. -- Antoine
Re: texlive & evince
Hi, On Thu, Feb 07, 2019 at 05:34:35PM +0100, Jeremie Courreges-Anglas wrote: > > Index: pkg/PLIST-synctex > > === > > RCS file: pkg/PLIST-synctex > > diff -N pkg/PLIST-synctex > > --- /dev/null 1 Jan 1970 00:00:00 - > > +++ pkg/PLIST-synctex 4 Feb 2019 17:39:55 - > > @@ -0,0 +1,15 @@ > > +@comment $OpenBSD$ > > +@conflict teTeX_base-* > > +@conflict pdfjam-* > > +@conflict ht-<2 > > +@conflict texlive_texmf-buildset-<2018 > > +@conflict texlive_base-<2018p1 > > The last line looks fine but I doubt that you need to copy the other > @conflict lines. Were the -synctex files once provided by one of those > packages? I looked at what espie@ did with the mktexlsr subpackage. The only conflict that might have conceivably included synctex was teTeX_base, but the synctex source code has copyright starting 2008, and teTeX died in 2006. So, I guess not. > As already discussed, I don't understand how it helps for those depends > to be that strict. Aren't files more commonly moved between > texlive/texmf subpackages, rather than from texlive/base to > texlive/texmf? Unless I'm missing something, we may easily avoid the > problem by only stating minimal versions: less chances for errors and less > churn for pkg_add -u. Please find a diff below. Yep, sounds good. Diff also looks good. So are we saying that, with that change, the diff I posted to move synctex into it's own sub-package would then be good to go (once we fix the merge conflicts that will be caused by your change)? > Well, as mentioned by kili there's also the option to force evince to > always use its internal copy. This means one more copy of synctex to > watch for security/portability issues, but on the other hand this would > decouple evince and texlive updates. Not a bad idea given the recent, > undocumented API/ABI break in libsynctex... > > https://github.com/jlaurens/synctex/issues/23 My gut feeling says we should use the one from TeX live, but I think we should hear from jasper@ and/or aja@, since they are the ones who maintain evince. Guys, what would you prefer? Thanks -- Best Regards Edd Barrett http://www.theunixzoo.co.uk
Re: texlive & evince
On Mon, Feb 04 2019, Edd Barrett wrote: > Hi, > > On Sat, Feb 02, 2019 at 12:25:55AM +0100, Matthias Kilian wrote: >> > - In tex live, put synctex into its own subpackage and LIB_DEPEND it. >> > - In tex live, move synctex into -mktexlsr. Evince already LIB_DEPENDS >> > this. >> > - Patch evince to always use its internal copy. >> > >> > FWIW, I'm leaning towards the first option. >> >> First or third. > > Alright, so here's a diff which makes the new -syntex subpackage in > texlive and another which adjusts evince to use it. See attached. > > I've tested `pkg_add -u` and all seems well. It seems the inequality > conflict constraint is essential, otherwise `pkg_add -u` isn't smart > enough to avoid collisions. Please check my work carefully. > Index: pkg/PLIST-synctex > === > RCS file: pkg/PLIST-synctex > diff -N pkg/PLIST-synctex > --- /dev/null 1 Jan 1970 00:00:00 - > +++ pkg/PLIST-synctex 4 Feb 2019 17:39:55 - > @@ -0,0 +1,15 @@ > +@comment $OpenBSD$ > +@conflict teTeX_base-* > +@conflict pdfjam-* > +@conflict ht-<2 > +@conflict texlive_texmf-buildset-<2018 > +@conflict texlive_base-<2018p1 The last line looks fine but I doubt that you need to copy the other @conflict lines. Were the -synctex files once provided by one of those packages? Note that you can't just put this in, you also need to bump some RUN_DEPENDS (and thus REVISION) in texlive/texmf/Makefile, else pkg_create will complain: --8<-- ===> Building package for texlive_texmf-docs-2018p0 Create /usr/ports/packages/amd64/no-arch/texlive_texmf-docs-2018p0.tgz Creating package texlive_texmf-docs-2018p0 checking dependencies|print/texlive/base,-main Error: Dependency texlive_base-2018p0 doesn't match FULLPKGNAME: texlive_base-2018p1 checksumming... Error: @depend print/texlive/base,-main:texlive_base-2018p0:texlive_base-2018p1 pattern texlive_base-2018p0 doesn't match default texlive_base-2018p1 Fatal error: can't continue at /usr/libdata/perl5/OpenBSD/PkgCreate.pm line 1675. *** Error 1 in . (/usr/ports/infrastructure/mk/bsd.port.mk:2027 '/usr/ports/packages/amd64/no-arch/texlive_texmf-docs-2018p0.tgz') *** Error 1 in . (/usr/ports/infrastructure/mk/bsd.port.mk:2488 '_internal-package') *** Error 1 in . (/usr/ports/infrastructure/mk/bsd.port.mk:2467 'package') *** Error 1 in /usr/ports/print/texlive/texmf (/usr/ports/infrastructure/mk/bsd.port.mk:3459 'repackage') -->8-- As already discussed, I don't understand how it helps for those depends to be that strict. Aren't files more commonly moved between texlive/texmf subpackages, rather than from texlive/base to texlive/texmf? Unless I'm missing something, we may easily avoid the problem by only stating minimal versions: less chances for errors and less churn for pkg_add -u. Please find a diff below. > As far as I can see, you cannot build evince without synctex, so even > the light flavor has to pull in synctex. Well, as mentioned by kili there's also the option to force evince to always use its internal copy. This means one more copy of synctex to watch for security/portability issues, but on the other hand this would decouple evince and texlive updates. Not a bad idea given the recent, undocumented API/ABI break in libsynctex... https://github.com/jlaurens/synctex/issues/23 Thoughts / ok? Index: Makefile === RCS file: /cvs/ports/print/texlive/texmf/Makefile,v retrieving revision 1.56 diff -u -p -r1.56 Makefile --- Makefile24 Jan 2019 16:38:41 - 1.56 +++ Makefile5 Feb 2019 03:04:38 - @@ -24,8 +24,9 @@ PKGNAME-docs =texlive_texmf-docs-${V} PKGNAME-context = texlive_texmf-context-${V} PKGNAME-buildset = texlive_texmf-buildset-${V} -REVISION-main =0 -REVISION-docs =0 +REVISION-main =1 +REVISION-docs =1 +REVISION-buildset =0 NO_BUILD = Yes NO_TEST = Yes @@ -42,16 +43,16 @@ MODRUBY_RUNDEP =No MODPY_BUILDDEP = No MODPY_RUNDEP = No -RUN_DEPENDS-main +=texlive_base-${V}p0:print/texlive/base,-main \ +RUN_DEPENDS-main +=texlive_base->=${V}p0:print/texlive/base,-main \ ${FULLPKGNAME-buildset}:print/texlive/texmf,-buildset RUN_DEPENDS-full = ${FULLPKGNAME-main}:print/texlive/texmf,-main \ ${MODPY_RUN_DEPENDS} \ ${_MODLUA_RUNDEPENDS} RUN_DEPENDS-context += ${FULLPKGNAME-full}:print/texlive/texmf,-full \ ${MODRUBY_RUNDEPENDS} -RUN_DEPENDS-docs +=texlive_base-${V}p0:print/texlive/base,-main \ +RUN_DEPENDS-docs +=texlive_base->=${V}p0:print/texlive/base,-main \ x11/p5-Tk -RUN_DEPENDS-buildset += texlive_mktexlsr-${V}:print/texlive/base,-mktexlsr +RUN_DEPENDS-buildset += texlive_mktexlsr->=${V}:print/texlive/base,-mktexlsr
[partially OT] ports and diffs and "smart" mail clients (was: texlive & evince)
Hi, On Mon, Feb 04, 2019 at 10:41:18PM +0100, Jeremie Courreges-Anglas wrote: > Date: Mon, 04 Feb 2019 22:41:18 +0100 > > Same diff from /usr/ports, easier to apply. > > Please ignore, this was just my mail client showing me the attachments > as inlined, without any hint. > > Learn something stupid everyday. Well, two things to learn: 1. When sending diffs for multiple ports in a single mail, base them all from /usr/ports (or at least from some common directory), to avoid confusion because of smart mail clients. (This was the not-so-offtopic part) 2. Mail clients hiding technical details (be it inlining of text attachments without giving a clear hint that they are attachments, be it hiding the real sender and recipient addresses and only displaying the comment part of the address) are a pest and should be patched to not do so by default. Ciao, Kili
Re: texlive & evince
On Mon, Feb 04 2019, Jeremie Courreges-Anglas wrote: [...] > Same diff from /usr/ports, easier to apply. Please ignore, this was just my mail client showing me the attachments as inlined, without any hint. Learn something stupid everyday. -- jca | PGP : 0x1524E7EE / 5135 92C1 AD36 5293 2BDF DDCC 0DFA 74AE 1524 E7EE
Re: texlive & evince
On Mon, Feb 04 2019, Edd Barrett wrote: > Hi, > > On Sat, Feb 02, 2019 at 12:25:55AM +0100, Matthias Kilian wrote: >> > - In tex live, put synctex into its own subpackage and LIB_DEPEND it. >> > - In tex live, move synctex into -mktexlsr. Evince already LIB_DEPENDS >> > this. >> > - Patch evince to always use its internal copy. >> > >> > FWIW, I'm leaning towards the first option. >> >> First or third. > > Alright, so here's a diff which makes the new -syntex subpackage in > texlive and another which adjusts evince to use it. See attached. > > I've tested `pkg_add -u` and all seems well. It seems the inequality > conflict constraint is essential, otherwise `pkg_add -u` isn't smart > enough to avoid collisions. Please check my work carefully. > > As far as I can see, you cannot build evince without synctex, so even > the light flavor has to pull in synctex. Same diff from /usr/ports, easier to apply. Index: graphics/evince/Makefile === RCS file: /cvs/ports/graphics/evince/Makefile,v retrieving revision 1.166 diff -u -p -r1.166 Makefile --- graphics/evince/Makefile6 Jan 2019 13:54:15 - 1.166 +++ graphics/evince/Makefile4 Feb 2019 21:19:29 - @@ -8,6 +8,8 @@ GNOME_VERSION= 3.30.2 MAJ_V= 3.0 SUBST_VARS=MAJ_V +REVISION = 0 + SHARED_LIBS += evdocument3 0.2 # 4.0 SHARED_LIBS += evview3 0.0 # 3.0 @@ -28,7 +30,7 @@ WANTLIB += gstreamer-1.0 gstvideo-1.0 gt WANTLIB += iconv intl jpeg lcms2 lzma m openjp2 orc-0.4 pango-1.0 WANTLIB += pangocairo-1.0 pangoft2-1.0 pcre pixman-1 png16 poppler WANTLIB += poppler-glib secret-1 tiff xcb xcb-render xcb-shm xml2 -WANTLIB += z aspell enchant-2 gspell-1 +WANTLIB += z aspell enchant-2 gspell-1 synctex COMPILER = base-clang ports-gcc base-gcc @@ -44,7 +46,8 @@ LIB_DEPENDS= multimedia/gstreamer1/plug print/poppler \ textproc/gspell \ x11/gnome/libsecret \ - x11/gtk+3 + x11/gtk+3 \ + print/texlive/base,-synctex RUN_DEPENDS= archivers/unrar \ archivers/unzip \ Index: print/texlive/base/Makefile === RCS file: /cvs/ports/print/texlive/base/Makefile,v retrieving revision 1.107 diff -u -p -r1.107 Makefile --- print/texlive/base/Makefile 24 Jan 2019 16:38:41 - 1.107 +++ print/texlive/base/Makefile 4 Feb 2019 21:19:29 - @@ -2,16 +2,18 @@ COMMENT-main = base binaries for TeXLive typesetting distribution COMMENT-mktexlsr = mktexlsr for TeXLive typesetting distribution +COMMENT-synctex = synctex for TeXLive typesetting distribution DISTNAME = texlive-${DIST_V}-source PKGNAME = texlive_base-${V} WRKDIST = ${WRKDIR}/texlive-${DIST_V}-source -REVISION-main =0 +REVISION-main =1 -MULTI_PACKAGES = -main -mktexlsr +MULTI_PACKAGES = -main -mktexlsr -synctex PKGNAME-mktexlsr = texlive_mktexlsr-${V} PKGNAME-main = ${PKGNAME} +PKGNAME-synctex = texlive_synctex-${V} DISTFILES =${DISTNAME}${EXTRACT_SUFX} \ texlive-${DIST_V}-extra${EXTRACT_SUFX} @@ -27,14 +29,15 @@ WANTLIB-main += Xmu Xpm Xrender Xt avahi WANTLIB-main += cups cupsimage dbus-1 expat ffi fontconfig freetype WANTLIB-main += gd glib-2.0 gmp gnutls graphite2 gs harfbuzz harfbuzz-icu WANTLIB-main += hogweed iconv icudata icui18n icuio icuuc idn2 ijs -WANTLIB-main += intl jbig2dec jpeg lcms2 m nettle p11-kit paper pcre -WANTLIB-main += pixman-1 png potrace tasn1 tiff unistring webp xcb +WANTLIB-main += intl jbig2dec jpeg lcms2 m nettle p11-kit paper +WANTLIB-main += pixman-1 png16 potrace tasn1 tiff unistring webp xcb WANTLIB-main += xcb-render xcb-shm z zzip - WANTLIB-main += kpathsea WANTLIB-mktexlsr = c +WANTLIB-synctex = c m z + RUN_DEPENDS-main = print/ghostscript/gnu \ print/detex \ print/dvi2tty \ @@ -42,7 +45,8 @@ RUN_DEPENDS-main =print/ghostscript/gnu print/psutils \ print/t1utils \ texlive_texmf-buildset-${V}:print/texlive/texmf,-buildset \ - devel/desktop-file-utils + devel/desktop-file-utils \ + ${BASE_PKGPATH},-synctex LIB_DEPENDS-main = converters/libiconv \ graphics/png \ Index: print/texlive/base/pkg/DESCR-synctex === RCS file: print/texlive/base/pkg/DESCR-synctex diff -N print/texlive/base/pkg/DESCR-synctex --- /dev/null 1 Jan 1970 00:00:00 - +++ print/texlive/base/pkg/DESCR-synctex4 Feb 2019
Re: texlive & evince
Hi, On Sat, Feb 02, 2019 at 12:25:55AM +0100, Matthias Kilian wrote: > > - In tex live, put synctex into its own subpackage and LIB_DEPEND it. > > - In tex live, move synctex into -mktexlsr. Evince already LIB_DEPENDS > > this. > > - Patch evince to always use its internal copy. > > > > FWIW, I'm leaning towards the first option. > > First or third. Alright, so here's a diff which makes the new -syntex subpackage in texlive and another which adjusts evince to use it. See attached. I've tested `pkg_add -u` and all seems well. It seems the inequality conflict constraint is essential, otherwise `pkg_add -u` isn't smart enough to avoid collisions. Please check my work carefully. As far as I can see, you cannot build evince without synctex, so even the light flavor has to pull in synctex. -- Best Regards Edd Barrett http://www.theunixzoo.co.uk Index: Makefile === RCS file: /cvs/ports/graphics/evince/Makefile,v retrieving revision 1.166 diff -u -p -r1.166 Makefile --- Makefile6 Jan 2019 13:54:15 - 1.166 +++ Makefile4 Feb 2019 20:17:07 - @@ -8,6 +8,8 @@ GNOME_VERSION= 3.30.2 MAJ_V= 3.0 SUBST_VARS=MAJ_V +REVISION = 0 + SHARED_LIBS += evdocument3 0.2 # 4.0 SHARED_LIBS += evview3 0.0 # 3.0 @@ -28,7 +30,7 @@ WANTLIB += gstreamer-1.0 gstvideo-1.0 gt WANTLIB += iconv intl jpeg lcms2 lzma m openjp2 orc-0.4 pango-1.0 WANTLIB += pangocairo-1.0 pangoft2-1.0 pcre pixman-1 png16 poppler WANTLIB += poppler-glib secret-1 tiff xcb xcb-render xcb-shm xml2 -WANTLIB += z aspell enchant-2 gspell-1 +WANTLIB += z aspell enchant-2 gspell-1 synctex COMPILER = base-clang ports-gcc base-gcc @@ -44,7 +46,8 @@ LIB_DEPENDS= multimedia/gstreamer1/plug print/poppler \ textproc/gspell \ x11/gnome/libsecret \ - x11/gtk+3 + x11/gtk+3 \ + print/texlive/base,-synctex RUN_DEPENDS= archivers/unrar \ archivers/unzip \ Index: Makefile === RCS file: /cvs/ports/print/texlive/base/Makefile,v retrieving revision 1.107 diff -u -p -r1.107 Makefile --- Makefile24 Jan 2019 16:38:41 - 1.107 +++ Makefile4 Feb 2019 16:40:55 - @@ -2,16 +2,18 @@ COMMENT-main = base binaries for TeXLive typesetting distribution COMMENT-mktexlsr = mktexlsr for TeXLive typesetting distribution +COMMENT-synctex = synctex for TeXLive typesetting distribution DISTNAME = texlive-${DIST_V}-source PKGNAME = texlive_base-${V} WRKDIST = ${WRKDIR}/texlive-${DIST_V}-source -REVISION-main =0 +REVISION-main =1 -MULTI_PACKAGES = -main -mktexlsr +MULTI_PACKAGES = -main -mktexlsr -synctex PKGNAME-mktexlsr = texlive_mktexlsr-${V} PKGNAME-main = ${PKGNAME} +PKGNAME-synctex = texlive_synctex-${V} DISTFILES =${DISTNAME}${EXTRACT_SUFX} \ texlive-${DIST_V}-extra${EXTRACT_SUFX} @@ -27,14 +29,15 @@ WANTLIB-main += Xmu Xpm Xrender Xt avahi WANTLIB-main += cups cupsimage dbus-1 expat ffi fontconfig freetype WANTLIB-main += gd glib-2.0 gmp gnutls graphite2 gs harfbuzz harfbuzz-icu WANTLIB-main += hogweed iconv icudata icui18n icuio icuuc idn2 ijs -WANTLIB-main += intl jbig2dec jpeg lcms2 m nettle p11-kit paper pcre -WANTLIB-main += pixman-1 png potrace tasn1 tiff unistring webp xcb +WANTLIB-main += intl jbig2dec jpeg lcms2 m nettle p11-kit paper +WANTLIB-main += pixman-1 png16 potrace tasn1 tiff unistring webp xcb WANTLIB-main += xcb-render xcb-shm z zzip - WANTLIB-main += kpathsea WANTLIB-mktexlsr = c +WANTLIB-synctex = c m z + RUN_DEPENDS-main = print/ghostscript/gnu \ print/detex \ print/dvi2tty \ @@ -42,7 +45,8 @@ RUN_DEPENDS-main =print/ghostscript/gnu print/psutils \ print/t1utils \ texlive_texmf-buildset-${V}:print/texlive/texmf,-buildset \ - devel/desktop-file-utils + devel/desktop-file-utils \ + ${BASE_PKGPATH},-synctex LIB_DEPENDS-main = converters/libiconv \ graphics/png \ Index: pkg/DESCR-synctex === RCS file: pkg/DESCR-synctex diff -N pkg/DESCR-synctex --- /dev/null 1 Jan 1970 00:00:00 - +++ pkg/DESCR-synctex 4 Feb 2019 20:10:30 - @@ -0,0 +1,6 @@ +TeX Live is an easy way to get up and running with TeX. It provides a +comprehensive TeX system which includes all the major TeX-related +programs, macro packages, and fonts that are free software, including
Re: texlive & evince
Hi, On Fri, Feb 01, 2019 at 10:49:56PM +, Edd Barrett wrote: > On Thu, Jan 24, 2019 at 10:07:15AM +, Edd Barrett wrote: > > I agree that it's not good for users to have to install a texmf subset > > just to use evince. I'm OK with evince lib depending upon texlive-base > > in the interim until we decide what to do, but in the long term we don't > > want that heavy dependency. > > So, unless I am mistaken, our evince package works because it either > picks up synctex from texlive_base if it's installed by chance at pkg > time, or else it is using its internal copy of synctex. Yes, that's bad. Please note that evince has a 'light' FLAVOR. which may make things even more complicated. > What do we want to do? > > - In tex live, put synctex into its own subpackage and LIB_DEPEND it. > - In tex live, move synctex into -mktexlsr. Evince already LIB_DEPENDS this. > - Patch evince to always use its internal copy. > > FWIW, I'm leaning towards the first option. First or third. The second option just sounds wrong -- mktexlsr is meant to only include, well, mktexlsr and the kpathsea bits, isn't it?. Has anyone build logs of evince,light around? Does it check for synctex, and build differently wether it's already installed or not? Ciao, Kili
Re: texlive & evince
Hi, On Thu, Jan 24, 2019 at 10:07:15AM +, Edd Barrett wrote: > I agree that it's not good for users to have to install a texmf subset > just to use evince. I'm OK with evince lib depending upon texlive-base > in the interim until we decide what to do, but in the long term we don't > want that heavy dependency. So, unless I am mistaken, our evince package works because it either picks up synctex from texlive_base if it's installed by chance at pkg time, or else it is using its internal copy of synctex. What do we want to do? - In tex live, put synctex into its own subpackage and LIB_DEPEND it. - In tex live, move synctex into -mktexlsr. Evince already LIB_DEPENDS this. - Patch evince to always use its internal copy. FWIW, I'm leaning towards the first option. Cheers! -- Best Regards Edd Barrett http://www.theunixzoo.co.uk
Re: texlive & evince
Hi, On Thu, Jan 24, 2019 at 03:31:41AM +0100, Jeremie Courreges-Anglas wrote: > While here I've noticed complains about an ABI break in libsynctex. > > Please find an updated diff below that also bumps shlib version of > libptexenc and libsynctex. Still ok? Ah, please feel free to bump all of the libraries. I usually do to be on the safe side, but it looks like I forgot this time :( About Kili's concern: > We still have to watch out for evince (the default flavor now > lib_depends on texlive/base,-main, and we still don't want more than > texlive/base,-mktexlsr for the light flavor). I agree that it's not good for users to have to install a texmf subset just to use evince. I'm OK with evince lib depending upon texlive-base in the interim until we decide what to do, but in the long term we don't want that heavy dependency. Whilst moving synctex into -mktexlsr is one solution, it kind of renders the name of the package odd. The current contents of -mktexlsr are not really all that related to libsynctex. I can think of a couple of other options: We could add a new -synctex subpackage, which contains the synctex includes and lib. However, we may end up in a scenario in the future where we have many little library subpackages, which will be annoying to maintain. We could instead put all of the bits external ports might like to use without depending on the texmf into -mktexlsr and rename the subpackage appropriately. I had initially though -libs might be a good name, but that's also misleading, as there are already binaries included in -mktexlsr (e.g. kpse* and mktexlsr itself). Maybe -shared or -tools? The latter option is more work upfront, as we'd have to sweep the tree updating dependencies, and perhaps add a quirk(?), but I think will be less effort to maintain in the long run. Thoughts? -- Best Regards Edd Barrett http://www.theunixzoo.co.uk
Re: texlive & evince
On Thu, Jan 24 2019, Jeremie Courreges-Anglas wrote: > On Wed, Jan 23 2019, Matthias Kilian wrote: >> Hi, >> >> On Wed, Jan 23, 2019 at 10:16:43PM +0100, Jeremie Courreges-Anglas wrote: >>> > First, this looks like synctex_version.h is missing from texlive_base, >>> > which should be fixed anyway. >>> >>> Here's a patch to address this, similar to what debian did: >>> >>> >>> https://salsa.debian.org/tex-team/texlive-bin/commit/7764b511543f2218527c068763827f2d5cd44ca5 >>> >>> I only patched the .in file though, to avoid an autoreconf dance. >>> >>> ok? >> >> Ok, if it packages for you. > > Well it does: > > $ pkg_info | grep texlive > texlive_base-2018p0 base binaries for TeXLive typesetting distribution > texlive_mktexlsr-2018 mktexlsr for TeXLive typesetting distribution > texlive_texmf-buildset-2018 smallest texlive texmf for building ports > texlive_texmf-minimal-2018p0 texlive texmf for basic functionality While here I've noticed complains about an ABI break in libsynctex. https://github.com/jlaurens/synctex/issues/23 Please find an updated diff below that also bumps shlib version of libptexenc and libsynctex. Still ok? Here's the check_sym output between tl2017 and tl2018 for the four libraries shipped by tl2017. --8<-- usr/local/lib/libkpathsea.so.8.1 --> /usr/local/lib/libkpathsea.so.8.1 No dynamic export changes usr/local/lib/libptexenc.so.5.1 --> /usr/local/lib/libptexenc.so.5.1 Dynamic export changes: added: setstdinenc usr/local/lib/libsynctex.so.1.0 --> /usr/local/lib/libsynctex.so.1.0 Dynamic export changes: added: _synctex_debug _synctex_free _synctex_log _synctex_node_next synctex_form synctex_form_content synctex_iterator_count synctex_iterator_free synctex_iterator_has_next synctex_iterator_new_display synctex_iterator_new_edit synctex_iterator_next_result synctex_iterator_reset synctex_node_arg_sibling synctex_node_depth synctex_node_free synctex_node_get_name synctex_node_hbox_depth synctex_node_hbox_h synctex_node_hbox_height synctex_node_hbox_v synctex_node_hbox_width synctex_node_height synctex_node_last_child synctex_node_last_sibling synctex_node_new synctex_node_parent_form synctex_node_parent_sheet synctex_node_target synctex_node_target_type synctex_node_visible_depth synctex_node_visible_height synctex_node_weight synctex_reader_free synctex_reader_init_with_output_file synctex_scanner_display_switcher synctex_scanner_input_with_tag synctex_scanner_new synctex_scanner_next_result synctex_scanner_reset_result synctex_scanner_set_display_switcher removed: _synctex_buffer_get_available_size _synctex_decode_int _synctex_decode_string _synctex_hbox_setup_visible _synctex_match_string _synctex_next_line _synctex_node_distance_to_point _synctex_open _synctex_point_h_distance _synctex_point_in_box _synctex_point_v_distance _synctex_scan_content _synctex_scan_float_and_dimension _synctex_scan_input _synctex_scan_named _synctex_scan_nested_sheet _synctex_scan_post_scriptum _synctex_scan_postamble _synctex_scan_preamble _synctex_scan_sheet _synctex_scanner_get_tag _synctex_setup_visible_box synctex_next_result synctex_node_charindex synctex_node_sheet External reference changes: added: __isthreaded exit feof fgetc fputs gzwrite snprintf vasprintf removed: gzprintf PLT added: _synctex_free synctex_iterator_new_display synctex_iterator_new_edit synctex_node_child synctex_node_hbox_h synctex_node_hbox_v synctex_reader_init_with_output_file synctex_scanner_new PLT removed: _synctex_buffer_get_available_size _synctex_decode_string _synctex_hbox_setup_visible _synctex_match_string _synctex_node_distance_to_point _synctex_open _synctex_point_h_distance _synctex_point_v_distance _synctex_scan_content _synctex_scan_float_and_dimension _synctex_scan_input _synctex_scan_nested_sheet _synctex_scan_post_scriptum _synctex_scan_postamble _synctex_scan_preamble _synctex_scan_sheet _synctex_scanner_get_tag _synctex_setup_visible_box usr/local/lib/libtexlua52.so.0.0 --> /usr/local/lib/libtexlua52.so.0.0 No dynamic export changes -->8-- Index: base/Makefile === RCS file: /cvs/ports/print/texlive/base/Makefile,v
Re: texlive & evince
On Wed, Jan 23 2019, Matthias Kilian wrote: > Hi, > > On Wed, Jan 23, 2019 at 10:16:43PM +0100, Jeremie Courreges-Anglas wrote: >> > First, this looks like synctex_version.h is missing from texlive_base, >> > which should be fixed anyway. >> >> Here's a patch to address this, similar to what debian did: >> >> >> https://salsa.debian.org/tex-team/texlive-bin/commit/7764b511543f2218527c068763827f2d5cd44ca5 >> >> I only patched the .in file though, to avoid an autoreconf dance. >> >> ok? > > Ok, if it packages for you. Well it does: $ pkg_info | grep texlive texlive_base-2018p0 base binaries for TeXLive typesetting distribution texlive_mktexlsr-2018 mktexlsr for TeXLive typesetting distribution texlive_texmf-buildset-2018 smallest texlive texmf for building ports texlive_texmf-minimal-2018p0 texlive texmf for basic functionality > We still have to watch out for evince > (the default flavor now lib_depends on texlive/base,-main, and we > still don't want more than texlive/base,-mktexlsr for the light > flavor). Indeed. Looks like there's no easy way to disable system synctex auto-detection. As a quick fix, evince could be patched to always build the internal copy. -- jca | PGP : 0x1524E7EE / 5135 92C1 AD36 5293 2BDF DDCC 0DFA 74AE 1524 E7EE
Re: texlive & evince
Hi, On Wed, Jan 23, 2019 at 10:16:43PM +0100, Jeremie Courreges-Anglas wrote: > > First, this looks like synctex_version.h is missing from texlive_base, > > which should be fixed anyway. > > Here's a patch to address this, similar to what debian did: > > > https://salsa.debian.org/tex-team/texlive-bin/commit/7764b511543f2218527c068763827f2d5cd44ca5 > > I only patched the .in file though, to avoid an autoreconf dance. > > ok? Ok, if it packages for you. We still have to watch out for evince (the default flavor now lib_depends on texlive/base,-main, and we still don't want more than texlive/base,-mktexlsr for the light flavor). Ciao, Kili
Re: texlive & evince
On Wed, Jan 23 2019, Matthias Kilian wrote: > Hi, > > On Wed, Jan 23, 2019 at 10:54:02AM +, Edd Barrett wrote: >> FWIW, I just did a manual "make build" in graphics/evince, which resulted in: >> >> ``` >> In file included from ev-document.c:29: >> /usr/local/include/synctex/synctex_parser.h:53:10: fatal error: >> 'synctex_version.h' file not found >> #include "synctex_version.h" >> ^~~ > > First, this looks like synctex_version.h is missing from texlive_base, > which should be fixed anyway. Here's a patch to address this, similar to what debian did: https://salsa.debian.org/tex-team/texlive-bin/commit/7764b511543f2218527c068763827f2d5cd44ca5 I only patched the .in file though, to avoid an autoreconf dance. ok? Index: base/Makefile === RCS file: /cvs/ports/print/texlive/base/Makefile,v retrieving revision 1.106 diff -u -p -r1.106 Makefile --- base/Makefile 20 Jan 2019 11:17:32 - 1.106 +++ base/Makefile 23 Jan 2019 21:14:53 - @@ -7,6 +7,8 @@ DISTNAME = texlive-${DIST_V}-source PKGNAME = texlive_base-${V} WRKDIST = ${WRKDIR}/texlive-${DIST_V}-source +REVISION-main =0 + MULTI_PACKAGES = -main -mktexlsr PKGNAME-mktexlsr = texlive_mktexlsr-${V} PKGNAME-main = ${PKGNAME} Index: base/patches/patch-texk_web2c_Makefile_in === RCS file: base/patches/patch-texk_web2c_Makefile_in diff -N base/patches/patch-texk_web2c_Makefile_in --- /dev/null 1 Jan 1970 00:00:00 - +++ base/patches/patch-texk_web2c_Makefile_in 23 Jan 2019 21:14:53 - @@ -0,0 +1,23 @@ +$OpenBSD$ + +Install missing header. + +Index: texk/web2c/Makefile.in +--- texk/web2c/Makefile.in.orig texk/web2c/Makefile.in +@@ -1762,6 +1762,7 @@ NROFF = nroff + MANS = $(dist_man_MANS) $(nodist_man_MANS) + DATA = $(pkgconfig_DATA) + am__syncinclude_HEADERS_DIST = synctexdir/synctex_parser.h \ ++ synctexdir/synctex_version.h \ + synctexdir/synctex_parser_utils.h + HEADERS = $(syncinclude_HEADERS) + RECURSIVE_CLEAN_TARGETS = mostlyclean-recursive clean-recursive \ +@@ -4021,6 +4022,7 @@ syncincludedir = ${includedir}/synctex + pkgconfigdir = ${libdir}/pkgconfig + @SYNCTEX_TRUE@syncinclude_HEADERS = \ + @SYNCTEX_TRUE@synctexdir/synctex_parser.h \ ++@SYNCTEX_TRUE@synctexdir/synctex_version.h \ + @SYNCTEX_TRUE@synctexdir/synctex_parser_utils.h + + @SYNCTEX_TRUE@pkgconfig_DATA = synctexdir/synctex.pc Index: base/pkg/PLIST-main === RCS file: /cvs/ports/print/texlive/base/pkg/PLIST-main,v retrieving revision 1.3 diff -u -p -r1.3 PLIST-main --- base/pkg/PLIST-main 20 Jan 2019 11:17:46 - 1.3 +++ base/pkg/PLIST-main 23 Jan 2019 21:14:53 - @@ -412,6 +412,7 @@ include/ptexenc/unicode.h include/synctex/ include/synctex/synctex_parser.h include/synctex/synctex_parser_utils.h +include/synctex/synctex_version.h include/texlua52/ include/texlua52/lauxlib.h include/texlua52/lua.h @@ -450,7 +451,6 @@ lib/pkgconfig/texlua53.pc ${P5SITE}/TeXLive/ ${P5SITE}/TeXLive/TLConfig.pm ${P5SITE}/TeXLive/TLUtils.pm -share/applications/ share/applications/xdvi.desktop share/doc/pkg-readmes/${PKGSTEM} @tag update-desktop-database Index: texmf/Makefile === RCS file: /cvs/ports/print/texlive/texmf/Makefile,v retrieving revision 1.55 diff -u -p -r1.55 Makefile --- texmf/Makefile 20 Jan 2019 11:17:48 - 1.55 +++ texmf/Makefile 23 Jan 2019 21:14:53 - @@ -24,6 +24,9 @@ PKGNAME-docs =texlive_texmf-docs-${V} PKGNAME-context = texlive_texmf-context-${V} PKGNAME-buildset = texlive_texmf-buildset-${V} +REVISION-main =0 +REVISION-docs =0 + NO_BUILD = Yes NO_TEST = Yes PKG_ARCH = * @@ -39,14 +42,14 @@ MODRUBY_RUNDEP =No MODPY_BUILDDEP = No MODPY_RUNDEP = No -RUN_DEPENDS-main +=texlive_base-${V}:print/texlive/base,-main \ +RUN_DEPENDS-main +=texlive_base-${V}p0:print/texlive/base,-main \ ${FULLPKGNAME-buildset}:print/texlive/texmf,-buildset RUN_DEPENDS-full = ${FULLPKGNAME-main}:print/texlive/texmf,-main \ ${MODPY_RUN_DEPENDS} \ ${_MODLUA_RUNDEPENDS} RUN_DEPENDS-context += ${FULLPKGNAME-full}:print/texlive/texmf,-full \ ${MODRUBY_RUNDEPENDS} -RUN_DEPENDS-docs +=texlive_base-${V}:print/texlive/base,-main \ +RUN_DEPENDS-docs +=texlive_base-${V}p0:print/texlive/base,-main \ x11/p5-Tk RUN_DEPENDS-buildset += texlive_mktexlsr-${V}:print/texlive/base,-mktexlsr -- jca | PGP : 0x1524E7EE / 5135 92C1 AD36 5293 2BDF DDCC 0DFA 74AE 1524 E7EE
Re: texlive & evince (was: CVS: cvs.openbsd.org: ports)
On Wed, Jan 23, 2019 at 08:11:28PM +0100, Matthias Kilian wrote: > Hi, > > On Wed, Jan 23, 2019 at 10:54:02AM +, Edd Barrett wrote: > > FWIW, I just did a manual "make build" in graphics/evince, which resulted > > in: > > > > ``` > > In file included from ev-document.c:29: > > /usr/local/include/synctex/synctex_parser.h:53:10: fatal error: > > 'synctex_version.h' file not found > > #include "synctex_version.h" > > ^~~ > > First, this looks like synctex_version.h is missing from texlive_base, > which should be fixed anyway. > > Second, the error happens if texlive_base is installed. If it's not > installed, evince uses it's own bunded version of synctex. Diff between > build log without and with texlive_base installed: > > --- build1.log Wed Jan 23 19:39:03 2019 > +++ build2.log Wed Jan 23 19:46:47 2019 > @@ -160,7 +160,7 @@ > checking for NAUTILUS... yes > checking for PREVIEWER... yes > checking for gobject-introspection... yes > -checking for SYNCTEX... no > +checking for SYNCTEX... yes > checking execinfo.h usability... no > checking execinfo.h presence... no > checking for execinfo.h... no > @@ -340,7 +340,7 @@ > Thumbnail cache ..: yes > Multimedia ...: yes > Spell Checker.: yes > -SyncTeX ..: internal > +SyncTeX ..: system > > I guess we don't want evince to lib-depend on texlive_base, but I didn't It already depends on print/texlive/base,-mktexlsr. For the !light FLAVOR. -- Antoine
Re: texlive & evince (was: CVS: cvs.openbsd.org: ports)
On Wed, Jan 23, 2019 at 08:11:28PM +0100, Matthias Kilian wrote: > I guess we don't want evince to lib-depend on texlive_base, but I didn't > find an autoconf knob for enforcing the internal synctex. Ignore that part. I forgot that evince already run-depends on texlive_base ;-) Ciao, Kili
Re: texlive & evince (was: CVS: cvs.openbsd.org: ports)
Hi, On Wed, Jan 23, 2019 at 10:54:02AM +, Edd Barrett wrote: > FWIW, I just did a manual "make build" in graphics/evince, which resulted in: > > ``` > In file included from ev-document.c:29: > /usr/local/include/synctex/synctex_parser.h:53:10: fatal error: > 'synctex_version.h' file not found > #include "synctex_version.h" > ^~~ First, this looks like synctex_version.h is missing from texlive_base, which should be fixed anyway. Second, the error happens if texlive_base is installed. If it's not installed, evince uses it's own bunded version of synctex. Diff between build log without and with texlive_base installed: --- build1.log Wed Jan 23 19:39:03 2019 +++ build2.log Wed Jan 23 19:46:47 2019 @@ -160,7 +160,7 @@ checking for NAUTILUS... yes checking for PREVIEWER... yes checking for gobject-introspection... yes -checking for SYNCTEX... no +checking for SYNCTEX... yes checking execinfo.h usability... no checking execinfo.h presence... no checking for execinfo.h... no @@ -340,7 +340,7 @@ Thumbnail cache ..: yes Multimedia ...: yes Spell Checker.: yes -SyncTeX ..: internal +SyncTeX ..: system I guess we don't want evince to lib-depend on texlive_base, but I didn't find an autoconf knob for enforcing the internal synctex. Ciao, Kili
Re: texlive & evince (was: CVS: cvs.openbsd.org: ports)
On Wed, Jan 23, 2019 at 01:38:30AM +0100, Antoine Jacoutot wrote: > Might be dpb junking? FWIW, I just did a manual "make build" in graphics/evince, which resulted in: ``` In file included from ev-document.c:29: /usr/local/include/synctex/synctex_parser.h:53:10: fatal error: 'synctex_version.h' file not found #include "synctex_version.h" ^~~ ``` -- Best Regards Edd Barrett http://www.theunixzoo.co.uk
Re: texlive & evince (was: CVS: cvs.openbsd.org: ports)
On Wed, Jan 23, 2019 at 12:43:18AM +0100, Matthias Kilian wrote: > Hi, >=20 > (putting this back to ports@, because i don't like discussions on > ports-changes@) >=20 > On Wed, Jan 23, 2019 at 12:01:12AM +0100, Antoine Jacoutot wrote: > > On Sun, Jan 20, 2019 at 04:17:54AM -0700, Edd Barrett wrote: > > > CVSROOT: /cvs > > > Module name: ports > > > Changes by: e...@cvs.openbsd.org2019/01/20 04:17:54 > [...] > > > Log message: > > > Update TeX Live to 2018. > >=20 > > I think this broke evince: > [...] > > cc -DHAVE_CONFIG_H -I. -I.. -DG_LOG_DOMAIN=3D"EvinceDocument" -DGNOMELO= CALEDIR=3D"/usr/local/share/locale" -DEV_BACKENDSDIR=3D"/usr/local/lib/evin= ce/4/backends" -DEV_BACKENDSBINARYVERSION=3D"4" -DEVINCE_COMPILATION -I/usr= /local/include -I/usr/local/include/gtk-3.0 -I/usr/local/include/at-spi2-at= k/2.0 -I/usr/local/include/at-spi-2.0 -I/usr/local/include/dbus-1.0 -I/usr/= local/lib/dbus-1.0/include -I/usr/local/include/gio-unix-2.0 -I/usr/X11R6/i= nclude -I/usr/local/include/cairo -I/usr/X11R6/include/libdrm -I/usr/local/= include/pango-1.0 -I/usr/local/include/harfbuzz -I/usr/local/include -I/usr= /local/include/fribidi -I/usr/local/include/atk-1.0 -I/usr/X11R6/include/pi= xman-1 -I/usr/X11R6/include/freetype2 -I/usr/local/include/libpng16 -I/usr/= local/include/gdk-pixbuf-2.0 -pthread -I/usr/local/include/glib-2.0 -I/usr/= local/lib/glib-2.0/include -I/usr/local/include/synctex -Wall -Wstrict-prot= otypes -Wnested-externs -Werror=3Dmissing-prototypes -Werror=3Dimplicit-fun= ction-declaration -Werror=3Dpointer-arith -Werror=3Dinit-self -Werror=3Dfor= mat-security -Werror=3Dformat=3D2 -Werror=3Dmissing-include-dirs -Werror=3D= return-type -O2 -pipe -Wno-error=3Dformat-nonliteral -MT libevdocument3_la-= ev-document.lo -MD -MP -MF .deps/libevdocument3_la-ev-document.Tpo -c ev-do= cument.c -fPIC -DPIC -o .libs/libevdocument3_la-ev-document.o > > In file included from ev-document.c:29: > > /usr/local/include/synctex/synctex_parser.h:53:10: fatal error: 'syncte= x_version.h' file not found > > #include "synctex_version.h" > > ^~~ > > 1 error generated. >=20 > This iw weird. When I tested edd's diff for texlive-2018, I also > removed all ports depending on it from my package repository, > including evince. And it did not fail, and I got proper evince > package (buit *after* texlive). Looking at recently built packages: Might be dpb junking? --=20 Antoine
Re: texlive & evince (was: CVS: cvs.openbsd.org: ports)
Hi, On Wed, Jan 23, 2019 at 12:43:18AM +0100, Matthias Kilian wrote: > This iw weird. When I tested edd's diff for texlive-2018, I also > removed all ports depending on it from my package repository, > including evince. And it did not fail, and I got proper evince > package (buit *after* texlive). I was also suprised that Kili's bulk didn't hit this. If you can reproduce the build failure, Kili, I'd say these upstream commits look like good candidates to try. They update the synctex parser that everyone seems to be copy-pasting into their projects these days. In order: https://gitlab.gnome.org/GNOME/evince/commit/ed2d42b8ed1496de951460cf0e9f4bb9f941695c https://gitlab.gnome.org/GNOME/evince/commit/dab0dc31dba530b7c5d92122cd2047e0f3f97452 https://gitlab.gnome.org/GNOME/evince/commit/678410e81d0c889f4db4e995ca451ed62b8a2eee (FWIW, there were some synctex changes between TL17 and TL18) -- Best Regards Edd Barrett http://www.theunixzoo.co.uk
texlive & evince (was: CVS: cvs.openbsd.org: ports)
Hi, (putting this back to ports@, because i don't like discussions on ports-changes@) On Wed, Jan 23, 2019 at 12:01:12AM +0100, Antoine Jacoutot wrote: > On Sun, Jan 20, 2019 at 04:17:54AM -0700, Edd Barrett wrote: > > CVSROOT:/cvs > > Module name:ports > > Changes by: e...@cvs.openbsd.org2019/01/20 04:17:54 [...] > > Log message: > > Update TeX Live to 2018. >=20 > I think this broke evince: [...] > cc -DHAVE_CONFIG_H -I. -I.. -DG_LOG_DOMAIN=3D"EvinceDocument" -DGNOMELOCA= LEDIR=3D"/usr/local/share/locale" -DEV_BACKENDSDIR=3D"/usr/local/lib/evince= /4/backends" -DEV_BACKENDSBINARYVERSION=3D"4" -DEVINCE_COMPILATION -I/usr/l= ocal/include -I/usr/local/include/gtk-3.0 -I/usr/local/include/at-spi2-atk/= 2.0 -I/usr/local/include/at-spi-2.0 -I/usr/local/include/dbus-1.0 -I/usr/lo= cal/lib/dbus-1.0/include -I/usr/local/include/gio-unix-2.0 -I/usr/X11R6/inc= lude -I/usr/local/include/cairo -I/usr/X11R6/include/libdrm -I/usr/local/in= clude/pango-1.0 -I/usr/local/include/harfbuzz -I/usr/local/include -I/usr/l= ocal/include/fribidi -I/usr/local/include/atk-1.0 -I/usr/X11R6/include/pixm= an-1 -I/usr/X11R6/include/freetype2 -I/usr/local/include/libpng16 -I/usr/lo= cal/include/gdk-pixbuf-2.0 -pthread -I/usr/local/include/glib-2.0 -I/usr/lo= cal/lib/glib-2.0/include -I/usr/local/include/synctex -Wall -Wstrict-protot= ypes -Wnested-externs -Werror=3Dmissing-prototypes -Werror=3Dimplicit-funct= ion-declaration -Werror=3Dpointer-arith -Werror=3Dinit-self -Werror=3Dforma= t-security -Werror=3Dformat=3D2 -Werror=3Dmissing-include-dirs -Werror=3Dre= turn-type -O2 -pipe -Wno-error=3Dformat-nonliteral -MT libevdocument3_la-ev= -document.lo -MD -MP -MF .deps/libevdocument3_la-ev-document.Tpo -c ev-docu= ment.c -fPIC -DPIC -o .libs/libevdocument3_la-ev-document.o > In file included from ev-document.c:29: > /usr/local/include/synctex/synctex_parser.h:53:10: fatal error: 'synctex_= version.h' file not found > #include "synctex_version.h" > ^~~ > 1 error generated. This iw weird. When I tested edd's diff for texlive-2018, I also removed all ports depending on it from my package repository, including evince. And it did not fail, and I got proper evince package (buit *after* texlive). Looking at recently built packages: -rw-rw-r-- 3 _pbuild _pbuild3702804 Jan 22 19:59 evince-3.30.2.tgz -rw-rw-r-- 3 _pbuild _pbuild3491721 Jan 21 22:14 evince-3.30.2-light.= tgz -rw-rw-r-- 3 _pbuild _pbuild 186028 Jan 20 04:46 texlive_mktexlsr-201= 8.tgz -rw-rw-r-- 3 _pbuild _pbuild 26209865 Jan 20 04:46 texlive_base-2018.tgz -rw-rw-r-- 4 _pbuild _pbuild 177361642 Jan 15 00:18 texlive_texmf-builds= et-2018.tgz -rw-rw-r-- 4 _pbuild _pbuild 270794739 Jan 15 00:16 texlive_texmf-minima= l-2018.tgz -rw-rw-r-- 4 _pbuild _pbuild 847869307 Jan 15 00:09 texlive_texmf-full-2= 018.tgz -rw-rw-r-- 4 _pbuild _pbuild 948758093 Jan 14 23:33 texlive_texmf-docs-2= 018.tgz -rw-rw-r-- 4 _pbuild _pbuild 12400397 Jan 14 23:27 texlive_texmf-contex= t-2018.tgz I'll try to rebuild evince on my (f*cking slow) build machine again and com= pare the build log against yours. Ciao, Kili