Re: intent to do a poppler transition
Andreas Barth píše v Út 10. 10. 2006 v 08:34 +0200: Hi, just to repeat: If you want to do the poppler transition, it is *your* task to check: * Steve Langasek ([EMAIL PROTECTED]) [061005 21:22]: - In any case, shouldn't we carefully check all affected packages, whether they FTBFS and whether they still work? This would IMO require a phase where all of them are in experimental, except poppler itself in case it gets a new dev package name. Yes, given the current release schedule, this new libpoppler transition will only be considered for etch if someone does rebuild and test all the reverse-dependencies, providing any necessary patches and documenting these to the release team. Given the current date, I don't expect the poppler transition to happen, unless that happens very fast. And even in that case, please speak with us before uploading the package to unstable. Considering my work schedule atm, let's postpone this transition to etch +1, so we have more time to do it right. Ondrej. -- Ondřej Surý [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: intent to do a poppler transition
* Ond??ej Surý ([EMAIL PROTECTED]) [061011 17:35]: Considering my work schedule atm, let's postpone this transition to etch +1, so we have more time to do it right. Ok, thanks. Cheers, Andi -- http://home.arcor.de/andreas-barth/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: intent to do a poppler transition
Hi, just to repeat: If you want to do the poppler transition, it is *your* task to check: * Steve Langasek ([EMAIL PROTECTED]) [061005 21:22]: - In any case, shouldn't we carefully check all affected packages, whether they FTBFS and whether they still work? This would IMO require a phase where all of them are in experimental, except poppler itself in case it gets a new dev package name. Yes, given the current release schedule, this new libpoppler transition will only be considered for etch if someone does rebuild and test all the reverse-dependencies, providing any necessary patches and documenting these to the release team. Given the current date, I don't expect the poppler transition to happen, unless that happens very fast. And even in that case, please speak with us before uploading the package to unstable. Cheers, Andi -- http://home.arcor.de/andreas-barth/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: intent to do a poppler transition
Frank Küster píše v Út 03. 10. 2006 v 16:29 +0200: Ondřej Surý [EMAIL PROTECTED] wrote: On Mon, 2006-10-02 at 22:39 +0200, Norbert Preining wrote: Dear Ondrej! Can you now tell us what the status is? It is a bit unclear for me? I can create new packages for texlive-bin with the changed patch, or leave it. Are the packages you want to upload to unstable already in experimental, or available in any other place? If yes I could at least try in my cowbuilder whether building works. 0.5.4-2 is in experimental (i386) and can be used as base for transition. Well, we can use them as a base for testing. However, it seems as if starting the transition would be a bit premature. I have seen a couple of questions that are not yet answered: - Since the API changed, shouldn't the -dev package change its name, or is this information in the Library Packaging Guide controversial? Or even if it's generally consensual, should the name still be kept unchanged because plain libpoppler doesn't guarantee any API anyway? - In any case, shouldn't we carefully check all affected packages, whether they FTBFS and whether they still work? This would IMO require a phase where all of them are in experimental, except poppler itself in case it gets a new dev package name. Step 1: Looks like ideal move would be to create libpoppler0.5-dev; -glib and -qt bindings didn't change API, so they could keep their name. Step 2: And I will introduce debian specific SONAME for libpoppler, so we are not hit by random ABI changes. Step 3: I would like to look at possibility of creating libpoppler-plain library with minimal subset of functionality needed by other packages and stable (incremental) API. Step 4: In future we should drop libpoppler-dev at all and have just -plain,-glib,-qt bindings -dev packages available. Agreed? Ondrej. -- Ondřej Surý [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: intent to do a poppler transition
Frank Küster píše v Čt 05. 10. 2006 v 13:00 +0200: Ondřej Surý [EMAIL PROTECTED] wrote: Step 1: Looks like ideal move would be to create libpoppler0.5-dev; -glib and -qt bindings didn't change API, so they could keep their name. Step 2: And I will introduce debian specific SONAME for libpoppler, so we are not hit by random ABI changes. I have no experience with that, but naively I would assume that we rather keep the soname until testing (or just information from upstream) indicates an actual ABI change? Sure, I just want to bump soname everytime ABI changes, not on every new upstream version. Kind regards, -- Ondřej Surý [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: intent to do a poppler transition
Ondřej Surý [EMAIL PROTECTED] wrote: Step 1: Looks like ideal move would be to create libpoppler0.5-dev; -glib and -qt bindings didn't change API, so they could keep their name. Step 2: And I will introduce debian specific SONAME for libpoppler, so we are not hit by random ABI changes. I have no experience with that, but naively I would assume that we rather keep the soname until testing (or just information from upstream) indicates an actual ABI change? Step 3: I would like to look at possibility of creating libpoppler-plain library with minimal subset of functionality needed by other packages and stable (incremental) API. Step 4: In future we should drop libpoppler-dev at all and have just -plain,-glib,-qt bindings -dev packages available. Agreed? Yes, especially if someone else does the work for steps 3 and 4... Regards, Frank -- Frank Küster Single Molecule Spectroscopy, Protein Folding @ Inst. f. Biochemie, Univ. Zürich Debian Developer (teTeX/TeXLive)
Re: intent to do a poppler transition
On Thu, Oct 05, 2006 at 12:01:50PM +0200, Ondřej Surý wrote: - Since the API changed, shouldn't the -dev package change its name, or is this information in the Library Packaging Guide controversial? Or even if it's generally consensual, should the name still be kept unchanged because plain libpoppler doesn't guarantee any API anyway? Step 1: Looks like ideal move would be to create libpoppler0.5-dev; -glib and -qt bindings didn't change API, so they could keep their name. Six packages build-depend on libpoppler-dev, but I understand that only one of them is affected by the API change; so it seems my concern about cost/benefit of changing the package name still applies here. Step 2: And I will introduce debian specific SONAME for libpoppler, so we are not hit by random ABI changes. Sounds good to me... Cheers, -- Steve Langasek Give me a lever long enough and a Free OS Debian Developer to set it on, and I can move the world. [EMAIL PROTECTED] http://www.debian.org/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: intent to do a poppler transition
On Tue, Oct 03, 2006 at 04:29:59PM +0200, Frank Küster wrote: 0.5.4-2 is in experimental (i386) and can be used as base for transition. Well, we can use them as a base for testing. However, it seems as if starting the transition would be a bit premature. I have seen a couple of questions that are not yet answered: - Since the API changed, shouldn't the -dev package change its name, or is this information in the Library Packaging Guide controversial? Yes, this is a very controversial recommendation in the library packaging guide. The recommendation there is to change the -dev package name for *any* API changes, no matter how small a subset of reverse-dependencies may be affected. If there are 9 packages build-depending on poppler, 8 of them can be binNMUed and one of them requires source changes to work with the new version, it's not an effective use of developer resources to impose a -dev package name change that will force maintainers of all 9 packages to make sourceful uploads. - In any case, shouldn't we carefully check all affected packages, whether they FTBFS and whether they still work? This would IMO require a phase where all of them are in experimental, except poppler itself in case it gets a new dev package name. Yes, given the current release schedule, this new libpoppler transition will only be considered for etch if someone does rebuild and test all the reverse-dependencies, providing any necessary patches and documenting these to the release team. This doesn't require uploading all of the packages to experimental; anyone wishing to work on this transition can do so in the environment of their choice and report the results to debian-release and the BTS. Thanks, -- Steve Langasek Give me a lever long enough and a Free OS Debian Developer to set it on, and I can move the world. [EMAIL PROTECTED] http://www.debian.org/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: intent to do a poppler transition
On 2006-10-05 16:48:19 -0400 Frank Küster [EMAIL PROTECTED] wrote: Two actually, tetex and texlive, since both build the same binary, pdftex. How do you know the others don't have a problem? Has anybody tried to build the others? Hmm... it looks like the new poppler does indeed change quite a few things. Neither PDFKit nor PopplerKit compiles with the libpoppler-dev that's currently in experimental, so that's at least four packages that don't compile out of the box. I have a patch ready for PDFKit, but there are a few changes that I'm not quite sure about. Ondřej, I'll send you an email privately recarding those changes. There seems to be a patch for PopplerKit in the Etoile SVN, but I haven't tried it yet. Ondřej Surý [EMAIL PROTECTED] wrote: Step 2: And I will introduce debian specific SONAME for libpoppler, so we are not hit by random ABI changes. Hmm... I'm not sure why you want to do this. Just looking at this current transition, and the fact that the ABI did indeed change in an incompatible manner, there's nothing to indicate to me that a new upstream version does not also require a new SONAME. But I don't know much about the history of poppler, so I'm just basing my opinion on the current situation. (Having a debian-specific SONAME for the glib or qt bindings might make sense, though, if the ABI did indeed not change for those libraries.) (And I'll again add my disclaimer about not knowing much about poppler to make a completely educated statement about it.) -- Hubert Chan - email Jabber: [EMAIL PROTECTED] - http://www.uhoreg.ca/ PGP/GnuPG key: 1024D/124B61FA (Key available at wwwkeys.pgp.net) Fingerprint: 96C5 012F 5F74 A5F7 1FF7 5291 AF29 C719 124B 61FA
Re: intent to do a poppler transition
P.S. I'm out of town and away from Internet access this weekend, until Tuesday. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: intent to do a poppler transition
On Mon, 2006-10-02 at 22:39 +0200, Norbert Preining wrote: Dear Ondrej! Can you now tell us what the status is? It is a bit unclear for me? I can create new packages for texlive-bin with the changed patch, or leave it. Are the packages you want to upload to unstable already in experimental, or available in any other place? If yes I could at least try in my cowbuilder whether building works. 0.5.4-2 is in experimental (i386) and can be used as base for transition. Ondrej. -- Ondřej Surý [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: intent to do a poppler transition
Ondřej Surý [EMAIL PROTECTED] wrote: On Mon, 2006-10-02 at 22:39 +0200, Norbert Preining wrote: Dear Ondrej! Can you now tell us what the status is? It is a bit unclear for me? I can create new packages for texlive-bin with the changed patch, or leave it. Are the packages you want to upload to unstable already in experimental, or available in any other place? If yes I could at least try in my cowbuilder whether building works. 0.5.4-2 is in experimental (i386) and can be used as base for transition. Well, we can use them as a base for testing. However, it seems as if starting the transition would be a bit premature. I have seen a couple of questions that are not yet answered: - Since the API changed, shouldn't the -dev package change its name, or is this information in the Library Packaging Guide controversial? Or even if it's generally consensual, should the name still be kept unchanged because plain libpoppler doesn't guarantee any API anyway? - In any case, shouldn't we carefully check all affected packages, whether they FTBFS and whether they still work? This would IMO require a phase where all of them are in experimental, except poppler itself in case it gets a new dev package name. Regards, Frank -- Frank Küster Single Molecule Spectroscopy, Protein Folding @ Inst. f. Biochemie, Univ. Zürich Debian Developer (teTeX/TeXLive)
Re: intent to do a poppler transition
Le lun 2 octobre 2006 14:26, Ondřej Surý a écrit : Hi, poppler 0.5.4 (f.d.o PDF rendering library) was declared stable by upstream and I would like to upload new version to unstable which changes SONAME from 0 to 1. No API changes were done between 0.4.x and 0.5.x so just rebuild with new libpoppler-dev should be enough. Also libpoppler.so moved to Libs.private:, so packages compiled against -qt,-glib binding should not directly depend on libpoppler after recompile (which is good thing :-). […] Through libpoppler0c2-qt: kdegraphics-kfile-plugins we will have a new dot release of kde soon, kde 3.5.5 that we'd like to push into etch, as it's (1) a minor fix-release (2) and that it address many of the kde current RC bugs. the 3.5.5 release is due to oct 10th, and I suppose the tarballs will be made available to us in the coming few days (probably end of the week). so we are ready to coordinate an upload here. -- ·O· Pierre Habouzit ··O[EMAIL PROTECTED] OOOhttp://www.madism.org pgp8rlCsWR4JL.pgp Description: PGP signature
Re: intent to do a poppler transition
On Mon, 2006-10-02 at 15:52 +0200, Frank Küster wrote: The following packages have unmet dependencies: libpoppler1: Depends: poppler-data but it is not installable E: Broken packages # apt-cache policy poppler-data poppler-data: Installed: (none) Candidate: (none) Version table: Just uploaded to experimental: Changes: poppler (0.5.4-2) experimental; urgency=low . * [debian/control]: poppler-data is non-free, do not depend on it (Closes: #389753) Ondrej -- Ondřej Surý [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: intent to do a poppler transition
On Mon, 2006-10-02 at 15:56 +0200, Norbert Preining wrote: On Mon, 02 Okt 2006, Ond??ej Surý wrote: poppler 0.5.4 (f.d.o PDF rendering library) was declared stable by upstream and I would like to upload new version to unstable which changes SONAME from 0 to 1. No API changes were done between 0.4.x and 0.5.x so just rebuild with new libpoppler-dev should be enough. Also I remember a patch in the bts that should make tetex/texlive ready for poppler 0.5, the patch came from the ubuntu people which are shipping 0.5. (http://patches.ubuntu.com/t/texlive-bin/texlive-bin_2005.dfsg.1-1ubuntu2.patch) So is there really ONLY a recompile necessary ... Duck and hides. Sorry, I was not aware of it. Anyway, I was going to suggest compiling against experimental package before upload to unstable happens, so we would find that out. Whole problem is more complicated. No external package should every never use libpopplerX directly (well at least according to upstream). According to upstream, it was never meant to be used that way. External packages should use either -glib or -qt bindings which have stable API and ABI. Main reason why I didn't do upload of 0.5.x series to unstable was possible ABI unstability (see upstream comments and possible solutins at https://bugs.freedesktop.org/show_bug.cgi?id=7054 ) Ondrej. -- Ondřej Surý [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: intent to do a poppler transition
On Mon, 02 Okt 2006, Ond??ej Surý wrote: poppler 0.5.4 (f.d.o PDF rendering library) was declared stable by upstream and I would like to upload new version to unstable which changes SONAME from 0 to 1. No API changes were done between 0.4.x and 0.5.x so just rebuild with new libpoppler-dev should be enough. Also I remember a patch in the bts that should make tetex/texlive ready for poppler 0.5, the patch came from the ubuntu people which are shipping 0.5. (http://patches.ubuntu.com/t/texlive-bin/texlive-bin_2005.dfsg.1-1ubuntu2.patch) So is there really ONLY a recompile necessary ... Best wishes Norbert --- Dr. Norbert Preining [EMAIL PROTECTED]Università di Siena Debian Developer [EMAIL PROTECTED] Debian TeX Group gpg DSA: 0x09C5B094 fp: 14DF 2E6C 0307 BE6D AD76 A9C0 D2BF 4AA3 09C5 B094 --- MARKET DEEPING (participial vb.) Stealing one piece of fruit from a street fruit-and- vegetable stall. --- Douglas Adams, The Meaning of Liff -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: intent to do a poppler transition
However, they are aware at least that other projects are interested in using poppler proper, and I think I told them we are already using it. We even talked about the possibility to provide a backend-free version with a clean API. AFAIR, the only reason for not providing it was that nobody had time to do it. I don't have an idea how this API should like, but I would be happy to join such project. (Better to help then deal with ABI breakages in libpoppler). It's completely inacceptable for pdftex to acquire a dependency on gtk or qt. It's just glib and not gtk, but I see your point. If using plain libpoppler turns out to be impossible, we'd rather switch back to using our embedded xpdf copy and have 10 security releases during each release cycle. I can always switch to debian SONAMEs if there is a need to, but simple library with clearly define API and no ABI breakages would be much better. Ondrej. -- Ondřej Surý [EMAIL PROTECTED] http://blog.rfc1925.org/ signature.asc Description: This is a digitally signed message part
Re: intent to do a poppler transition
Ondřej Surý [EMAIL PROTECTED] wrote: Gnome Team will take care of evince, and I am Ccing maintainers of affected packages (although I think that binary NMU can fix that?). Hm, while trying to check tetex-bin: # apt-get -t experimental install libpoppler1 libpoppler-dev Reading package lists... Done Building dependency tree... Done Some packages could not be installed. This may mean that you have requested an impossible situation or if you are using the unstable distribution that some required packages have not yet been created or been moved out of Incoming. The following information may help to resolve the situation: The following packages have unmet dependencies: libpoppler1: Depends: poppler-data but it is not installable E: Broken packages # apt-cache policy poppler-data poppler-data: Installed: (none) Candidate: (none) Version table: The reason for this is that poppler 0.4.x is very buggy and it will not be supported from upstream. Also number of affected packages in not very high. I generally support this move. Regards, Frank -- Frank Küster Single Molecule Spectroscopy, Protein Folding @ Inst. f. Biochemie, Univ. Zürich Debian Developer (teTeX/TeXLive)
Re: intent to do a poppler transition
Ondřej Surý [EMAIL PROTECTED] wrote: Hi, poppler 0.5.4 (f.d.o PDF rendering library) was declared stable by upstream and I would like to upload new version to unstable which changes SONAME from 0 to 1. No API changes were done between 0.4.x and 0.5.x so just rebuild with new libpoppler-dev should be enough. This is not correct, see for example http://bugs.debian.org/356079 Regards, Frank -- Frank Küster Single Molecule Spectroscopy, Protein Folding @ Inst. f. Biochemie, Univ. Zürich Debian Developer (teTeX/TeXLive)
Re: intent to do a poppler transition
Ondřej Surý [EMAIL PROTECTED] wrote: Whole problem is more complicated. No external package should every never use libpopplerX directly (well at least according to upstream). According to upstream, it was never meant to be used that way. External packages should use either -glib or -qt bindings which have stable API and ABI. (I'd rather say have a well-defined API at all, which xpdf and hence libpoppler hasn't) However, they are aware at least that other projects are interested in using poppler proper, and I think I told them we are already using it. We even talked about the possibility to provide a backend-free version with a clean API. AFAIR, the only reason for not providing it was that nobody had time to do it. It's completely inacceptable for pdftex to acquire a dependency on gtk or qt. If using plain libpoppler turns out to be impossible, we'd rather switch back to using our embedded xpdf copy and have 10 security releases during each release cycle. Regards, Frank -- Frank Küster Single Molecule Spectroscopy, Protein Folding @ Inst. f. Biochemie, Univ. Zürich Debian Developer (teTeX/TeXLive)
Re: intent to do a poppler transition
On 2006-10-02, Norbert Preining [EMAIL PROTECTED] wrote: On Mon, 02 Okt 2006, Ond??ej Surý wrote: poppler 0.5.4 (f.d.o PDF rendering library) was declared stable by upstream and I would like to upload new version to unstable which changes SONAME from 0 to 1. No API changes were done between 0.4.x and 0.5.x so just rebuild with new libpoppler-dev should be enough. Also I remember a patch in the bts that should make tetex/texlive ready for poppler 0.5, the patch came from the ubuntu people which are shipping 0.5. (http://patches.ubuntu.com/t/texlive-bin/texlive-bin_2005.dfsg.1-1ubuntu2.patch) *Sigh* Is this another embedded instance of the xpdf code? Please have it fixed before Etch, the Sarge situation has been a complete disaster. Cheers, Moritz -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: intent to do a poppler transition
Frank Küster wrote: It's completely inacceptable for pdftex to acquire a dependency on gtk or qt. If using plain libpoppler turns out to be impossible, we'd Nearly 1000 packages in sid depend on glib, it's not that it's a completely obscure lib causing major problems. Please elaborate why that is completely unacceptable. rather switch back to using our embedded xpdf copy and have 10 security releases during each release cycle. Great, just unload the work onto someone else! -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: intent to do a poppler transition
On 2006-10-02 08:26:08 -0400 Ondřej Surý [EMAIL PROTECTED] wrote: Affected packages (directly): libpopplerkit0 libpdfkit0 FYI, these two packages are also part of the (currently ongoing) GNUstep library transition. I'll check to see they will still compile with the new poppler (though I probably won't be able to get around to it until a couple of days). -- Hubert Chan - email Jabber: [EMAIL PROTECTED] - http://www.uhoreg.ca/ PGP/GnuPG key: 1024D/124B61FA (Key available at wwwkeys.pgp.net) Fingerprint: 96C5 012F 5F74 A5F7 1FF7 5291 AF29 C719 124B 61FA
Re: intent to do a poppler transition
On 2006-10-02 09:56:50 -0400 Norbert Preining [EMAIL PROTECTED] wrote: On Mon, 02 Okt 2006, Ond??ej Surý wrote: poppler 0.5.4 (f.d.o PDF rendering library) was declared stable by upstream and I would like to upload new version to unstable which changes SONAME from 0 to 1. No API changes were done between 0.4.x and 0.5.x so just rebuild with new libpoppler-dev should be enough. Also I remember a patch in the bts that should make tetex/texlive ready for poppler 0.5, the patch came from the ubuntu people which are shipping 0.5. (http://patches.ubuntu.com/t/texlive-bin/texlive-bin_2005.dfsg.1-1ubuntu2.patch) Hmm... If the API has changed, then shouldn't the -dev package name be bumped? -- Hubert Chan - email Jabber: [EMAIL PROTECTED] - http://www.uhoreg.ca/ PGP/GnuPG key: 1024D/124B61FA (Key available at wwwkeys.pgp.net) Fingerprint: 96C5 012F 5F74 A5F7 1FF7 5291 AF29 C719 124B 61FA
Re: intent to do a poppler transition
Dear Ondrej! Can you now tell us what the status is? It is a bit unclear for me? I can create new packages for texlive-bin with the changed patch, or leave it. Are the packages you want to upload to unstable already in experimental, or available in any other place? If yes I could at least try in my cowbuilder whether building works. Best wishes Norbert --- Dr. Norbert Preining [EMAIL PROTECTED]Università di Siena Debian Developer [EMAIL PROTECTED] Debian TeX Group gpg DSA: 0x09C5B094 fp: 14DF 2E6C 0307 BE6D AD76 A9C0 D2BF 4AA3 09C5 B094 --- DALRYMPLE (n.) Dalarymples are the things you pay extra for on pieces of hand-made craftwork - the rough edges, the paint smudges and the holes in the glazing. --- Douglas Adams, The Meaning of Liff -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]