On 11/30/2010 01:42 AM, Jakub Hrozek wrote: > > Utilizes PO4A to extract translatable strings from Docbook XML sources > > and allows translators to submit ordinary .PO files. PO4A then generates > > translated Docbook documents that can be used to generate translated end > > user documentation. > > > > https://fedorahosted.org/sssd/ticket/297 > > > > The second patch adds a skeleton of Czech translation I used for testing > > -- at least one man page is localized so the whole process is testable. > > > > The general approach is based on what some of the Debian-based projects do. > > > > I have two questions about the patch: > > > > * the way the patch is written now, the localized manpages (and the > > respective xlm sources) are generated during every "make" invocation -- > > so they are always in sync. I couldn't decide it makes more sense to add > > a specialized "docs" and "docsinstall" targets or something > > similar..which would be trivial to do, I'm just not sure whether it's > > better or worse than keeping things done with "make all". > > Please use the existing PO files as a guideline. We only regenerate the .po files during a "make dist", because we don't want them to constantly be cluttering our git status.

Please make the new targets for updating the po files as a dependency on the update-pot target in po/Makefile if possible.

This way, they're only generated during "make dist". This will be behaving the same way that the current translatable files do.

It will be the release engineer's responsibility to build these translated manpages and commit them at the string freeze.

"make install" should just use whatever previously-generated PO files are available. > > * the .po files are distributed, not the localized .xml sources. I > > think it makes more sense since the localized .xml sources are generated > > from the .po files and the original .xml sources anyway, but it would be > > easy to change the distribution to .xml and only regenerate the .xml > > sources when releasing a new upstream > > tarball. We need to make sure the po files are distributed, otherwise Transifex won't work.

-----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.11 (GNU/Linux) Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org/ iEYEARECAAYFAkz2OjwACgkQHsardTLnvCW+vwCg3nX4sQbon9bq3tq/REheFeSq kcwAnRZ//laBsVl1Ch+1zU6SaHuRzq2L =RFfb -----END PGP SIGNATURE-----

On 12/01/2010 07:06 AM, Jakub Hrozek wrote: > On 11/30/2010 12:51 PM, Stephen Gallagher wrote: >> On 11/30/2010 01:42 AM, Jakub Hrozek wrote: >>>> Utilizes PO4A to extract translatable strings from Docbook XML sources >>>> and allows translators to submit ordinary .PO files. PO4A then generates >>>> translated Docbook documents that can be used to generate translated end >>>> user documentation. >>>> >>>> https://fedorahosted.org/sssd/ticket/297 >>>> >>>> The second patch adds a skeleton of Czech translation I used for testing >>>> -- at least one man page is localized so the whole process is testable. >>>> >>>> The general approach is based on what some of the Debian-based projects do. >>>> >>>> I have two questions about the patch: >>>> >>>> * the way the patch is written now, the localized manpages (and the >>>> respective xlm sources) are generated during every "make" invocation -- >>>> so they are always in sync. I couldn't decide it makes more sense to add >>>> a specialized "docs" and "docsinstall" targets or something >>>> similar..which would be trivial to do, I'm just not sure whether it's >>>> better or worse than keeping things done with "make all". >>>> >> Please use the existing PO files as a guideline. We only regenerate the >> .po files during a "make dist", because we don't want them to constantly >> be cluttering our git status. > OK, done >> Please make the new targets for updating the po files as a dependency on >> the update-pot target in po/Makefile if possible. > Done >> This way, they're only >> generated during "make dist". This will be behaving the same way that >> the current translatable files do. > The difference is that there are actually two generated files - the > translated XML files and the actual manpages. The translated XML files > are kind of middle step, because we can for instance decide to build > translated HTML or PDF documentation. See below. >> It will be the release engineer's responsibility to build these >> translated manpages and commit them at the string freeze. > I agree that this should be done for .PO files. But I think that the > translated XML files should be generated from .PO translations and the > original XML files during build. Because all information about the > structure of the documents is in the original XML files and translations > are in .PO files, the translatex .XML files are redundant and only .PO > files need to be distributed. > The attached patch does that and has a special "make > translated-manpages" target that is called in the specfile. I would > probably prefer to name it "make docs" and have the doxygen target be > called "make apidocs", but that's a detail. >> "make install" should just use whatever previously-generated PO files >> are available. >>>> * the .po files are distributed, not the localized .xml sources. I >>>> think it makes more sense since the localized .xml sources are generated >>>> from the .po files and the original .xml sources anyway, but it would be >>>> easy to change the distribution to .xml and only regenerate the .xml >>>> sources when releasing a new upstream >>>> tarball. >> We need to make sure the po files are distributed, otherwise Transifex >> won't work. Nack. Doesn't compile. Looks like you forgot to include LINGUAS.

-----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.11 (GNU/Linux) Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org/ iEYEARECAAYFAkz3rF0ACgkQeiVVYja6o6O7EACfahiauUF9Ev3jM2YZTu2b1qzP i78An3W9Nb3juA8yiSZkIwQ/VEvjBpjI =oZ9f -----END PGP SIGNATURE-----

On 12/14/2010 07:24 PM, Stephen Gallagher wrote: > I think you sent the wrong patch. I'm still seeing the 'make distcheck > failure' Sorry, you're right. Attached is a new patch, which Sumit helped me fix. As discussed on IRC, the for loop in the Makefile is not really in Makefile style. We might want to fix that so that the translated xml files are regenerated only when needed, but I think this can be tracked by a separate ticket. Also, I'm opening a ticket with po4a upstream to provide an option to po4a so it can do the job of po4a-translate but leverage the config file.

-----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.11 (GNU/Linux) Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org/ iEYEARECAAYFAk0JKZ0ACgkQeiVVYja6o6NwaQCfewB4cf3F9pMono+TaUEiMQtX iRcAn2+zJcqASAvo1lt96pc4CrXHQfaU =wb1o -----END PGP SIGNATURE-----

On 12/15/2010 09:48 PM, Stephen Gallagher wrote: > Nack. These patches fail to build when configure is run in a directory > other than the root of the source checkout (aka a parallel build directory) > See attached build log attempting to build from /dev/shm/test-build. Thanks, I missed this case in my testing (or rather, I though that make distcheck does essentially the same thing). Attached is a patch that always builds the included xml files as the real xml sources depend on them.

-----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.11 (GNU/Linux) Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org/ iEYEARECAAYFAk0KBrgACgkQeiVVYja6o6NVdACgmVBZlFUyYtaOPHrI0u7LYBmh yS8An1fHA+GjVKA2kOWPlAKTgUWF1A7s =4pfu -----END PGP SIGNATURE-----

On 12/16/2010 01:31 PM, Stephen Gallagher wrote: > On 12/16/2010 06:36 AM, Jakub Hrozek wrote: >> On 12/15/2010 09:48 PM, Stephen Gallagher wrote: >>> Nack. These patches fail to build when configure is run in a directory >>> other than the root of the source checkout (aka a parallel build directory) >>> See attached build log attempting to build from /dev/shm/test-build. >> Thanks, I missed this case in my testing (or rather, I though that make >> distcheck does essentially the same thing). >> Attached is a patch that always builds the included xml files as the >> real xml sources depend on them. > Nack. Running "make translated-manpages" now works in a parallel build > dir, but running "make distcheck" still fails. > Logs attached. > distcheck-clean.log shows the attempt to run "make distcheck" from a > perfectly clean checkout with this patch (patch applied, then 'git reset > --hard' to make sure there are no pre-generated translation files) > distcheck-translated-manpages shows the attempt to run "make distcheck" > from a checkout that was configured and then had "make > translated-manpages" run on it. > Both of these were run from a parallel build dir. Sumit managed to fix this issue. Attached is my previous patch with his fix.

-----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.11 (GNU/Linux) Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org/ iEYEARECAAYFAk0LcYwACgkQeiVVYja6o6NYYwCfY2J60KRcT5VNHvd88cjAgMjW XnMAn0qjI7QBkZsakuYXQ1sIKJarycwb =dqh8 -----END PGP SIGNATURE-----

On Fri, Dec 17, 2010 at 09:19:56AM -0500, Stephen Gallagher wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > On 12/16/2010 08:51 AM, Jakub Hrozek wrote: > > On 12/16/2010 01:31 PM, Stephen Gallagher wrote: > >> On 12/16/2010 06:36 AM, Jakub Hrozek wrote: > >>> On 12/15/2010 09:48 PM, Stephen Gallagher wrote: > >>>> Nack. These patches fail to build when configure is run in a directory > >>>> other than the root of the source checkout (aka a parallel build directory) > > > >>>> See attached build log attempting to build from /dev/shm/test-build. > > > > > >>> Thanks, I missed this case in my testing (or rather, I though that make > >>> distcheck does essentially the same thing). > > > >>> Attached is a patch that always builds the included xml files as the > >>> real xml sources depend on them. > > > > > >> Nack. Running "make translated-manpages" now works in a parallel build > >> dir, but running "make distcheck" still fails. > > > >> Logs attached. > > > >> distcheck-clean.log shows the attempt to run "make distcheck" from a > >> perfectly clean checkout with this patch (patch applied, then 'git reset > >> --hard' to make sure there are no pre-generated translation files) > > > >> distcheck-translated-manpages shows the attempt to run "make distcheck" > >> from a checkout that was configured and then had "make > >> translated-manpages" run on it. > > > >> Both of these were run from a parallel build dir. > > > > > > > > Sumit managed to fix this issue. Attached is my previous patch with his fix. > > > > Nack. Patch does not update translations when "make distdir" is run. > This is needed for ensuring that releases always have the latest > translation data available. > The attached patch should fix it, please apply it on top of the other two. bye, Sumit

> > - -- > Stephen Gallagher > RHCE 804006346421761 > > Delivering value year after year. > Red Hat ranks #1 in value among software vendors. > http://www.redhat.com/promo/vendor/ > _______________________________________________ > sssd-devel mailing list > sssd-devel(a)lists.fedorahosted.org > https://fedorahosted.org/mailman/listinfo/sssd-devel

On Mon, Dec 20, 2010 at 02:41:18PM +0100, Sumit Bose wrote: > On Mon, Dec 20, 2010 at 01:46:49PM +0100, Sumit Bose wrote: > > On Fri, Dec 17, 2010 at 09:19:56AM -0500, Stephen Gallagher wrote: > > > -----BEGIN PGP SIGNED MESSAGE----- > > > Hash: SHA1 > > > > > > On 12/16/2010 08:51 AM, Jakub Hrozek wrote: > > > > On 12/16/2010 01:31 PM, Stephen Gallagher wrote: > > > >> On 12/16/2010 06:36 AM, Jakub Hrozek wrote: > > > >>> On 12/15/2010 09:48 PM, Stephen Gallagher wrote: > > > >>>> Nack. These patches fail to build when configure is run in a directory > > > >>>> other than the root of the source checkout (aka a parallel build directory) > > > > > > > >>>> See attached build log attempting to build from /dev/shm/test-build. > > > > > > > > > > > >>> Thanks, I missed this case in my testing (or rather, I though that make > > > >>> distcheck does essentially the same thing). > > > > > > > >>> Attached is a patch that always builds the included xml files as the > > > >>> real xml sources depend on them. > > > > > > > > > > > >> Nack. Running "make translated-manpages" now works in a parallel build > > > >> dir, but running "make distcheck" still fails. > > > > > > > >> Logs attached. > > > > > > > >> distcheck-clean.log shows the attempt to run "make distcheck" from a > > > >> perfectly clean checkout with this patch (patch applied, then 'git reset > > > >> --hard' to make sure there are no pre-generated translation files) > > > > > > > >> distcheck-translated-manpages shows the attempt to run "make distcheck" > > > >> from a checkout that was configured and then had "make > > > >> translated-manpages" run on it. > > > > > > > >> Both of these were run from a parallel build dir. > > > > > > > > > > > > > > > > Sumit managed to fix this issue. Attached is my previous patch with his fix. > > > > > > > > > > > > Nack. Patch does not update translations when "make distdir" is run. > > > This is needed for ensuring that releases always have the latest > > > translation data available. > > > > > > > The attached patch should fix it, please apply it on top of the other > > two. > > > > bye, > > Sumit > > Here is a different version of the patch with the distdir-hook target > instead of dist-hook. > > bye, > Sumit > Another new version which can create a missing po directory. bye, Sumit

> > > > > > > > - -- > > > Stephen Gallagher > > > RHCE 804006346421761 > > > > > > Delivering value year after year. > > > Red Hat ranks #1 in value among software vendors. > > > http://www.redhat.com/promo/vendor/ > > > _______________________________________________ > > > sssd-devel mailing list > > > sssd-devel(a)lists.fedorahosted.org > > > https://fedorahosted.org/mailman/listinfo/sssd-devel

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 12/20/2010 11:03 AM, Sumit Bose wrote: > On Mon, Dec 20, 2010 at 04:13:10PM +0100, Sumit Bose wrote: >> On Mon, Dec 20, 2010 at 02:41:18PM +0100, Sumit Bose wrote: >>> On Mon, Dec 20, 2010 at 01:46:49PM +0100, Sumit Bose wrote: >>>> On Fri, Dec 17, 2010 at 09:19:56AM -0500, Stephen Gallagher wrote: > On 12/16/2010 08:51 AM, Jakub Hrozek wrote: >>>>>>> On 12/16/2010 01:31 PM, Stephen Gallagher wrote: >>>>>>>> On 12/16/2010 06:36 AM, Jakub Hrozek wrote: >>>>>>>>> On 12/15/2010 09:48 PM, Stephen Gallagher wrote:

Sorry, this is still a Nack.

On Mon, Dec 20, 2010 at 11:52:18AM -0500, Stephen Gallagher wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > On 12/20/2010 11:03 AM, Sumit Bose wrote: >> On Mon, Dec 20, 2010 at 04:13:10PM +0100, Sumit Bose wrote: >>> On Mon, Dec 20, 2010 at 02:41:18PM +0100, Sumit Bose wrote: >>>> On Mon, Dec 20, 2010 at 01:46:49PM +0100, Sumit Bose wrote: >>>>> On Fri, Dec 17, 2010 at 09:19:56AM -0500, Stephen Gallagher >>>>> wrote: >> On 12/16/2010 08:51 AM, Jakub Hrozek wrote: >>>>>>>> On 12/16/2010 01:31 PM, Stephen Gallagher wrote: >>>>>>>>> On 12/16/2010 06:36 AM, Jakub Hrozek wrote: >>>>>>>>>> On 12/15/2010 09:48 PM, Stephen Gallagher wrote: > .... > Sorry, this is still a Nack. .... ok, kind of a new start. We discussed that we want to have the translated xml files in the tar ball so that po4a is not needed to build translated man pages. I also choose to include the po4a config file into the repository instead of auto-generating it. The main reason, besides making the makefile simpler, is that we might want special options like char set for different languages.

-----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.11 (GNU/Linux) Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org/ iEYEARECAAYFAk0SLLEACgkQeiVVYja6o6OtcwCgmRtqQmwXH8yOUmLf9nOhdZiD RLQAniPdu3RYxTtAyMEbAdLCgys9Cg2m =NXDB -----END PGP SIGNATURE-----

On Wed, Dec 22, 2010 at 11:52:01AM -0500, Stephen Gallagher wrote: On 12/22/2010 10:36 AM, Sumit Bose wrote: >>> On Mon, Dec 20, 2010 at 11:52:18AM -0500, Stephen Gallagher wrote: >>>> -----BEGIN PGP SIGNED MESSAGE----- >>>> Hash: SHA1 >>>> >>>> On 12/20/2010 11:03 AM, Sumit Bose wrote: >>>>> On Mon, Dec 20, 2010 at 04:13:10PM +0100, Sumit Bose wrote: >>>>>> On Mon, Dec 20, 2010 at 02:41:18PM +0100, Sumit Bose wrote: >>>>>>> On Mon, Dec 20, 2010 at 01:46:49PM +0100, Sumit Bose wrote: >>>>>>>> On Fri, Dec 17, 2010 at 09:19:56AM -0500, Stephen Gallagher >>>>>>>> wrote: >>>>> On 12/16/2010 08:51 AM, Jakub Hrozek wrote: >>>>>>>>>>> On 12/16/2010 01:31 PM, Stephen Gallagher wrote: >>>>>>>>>>>> On 12/16/2010 06:36 AM, Jakub Hrozek wrote: >>>>>>>>>>>>> On 12/15/2010 09:48 PM, Stephen Gallagher wrote: >>>> >>> .... >>>> Sorry, this is still a Nack. >>> .... >>> >>> ok, kind of a new start. We discussed that we want to have the >>> translated xml files in the tar ball so that po4a is not needed to build >>> translated man pages. I also choose to include the po4a config file into >>> the repository instead of auto-generating it. The main reason, besides >>> making the makefile simpler, is that we might want special options like >>> char set for different languages. Nack. While 'make all' doesn't update the .pot files on manpage changes (correctly), doing a 'make clean' causes the .pot files to be regenerated. > New version attached. > bye, > Sumit All of my other tests seemed to be working smoothly, so hopefully this is the last iteration.

-----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.11 (GNU/Linux) Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org/ iEYEARECAAYFAk0SPdwACgkQeiVVYja6o6OAXQCfcLOvigYiI/MDSnRijQLnbUbS xKwAnRaPt9yq/h7zfMFvyg4keky/YXsH =vmOA -----END PGP SIGNATURE-----