-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
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".
* 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.
Thanks, Jakub
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
I haven't reviewed or tested this patch yet, but I have some comments on the summary. See below.
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.
- -- Stephen Gallagher RHCE 804006346421761
Delivering value year after year. Red Hat ranks #1 in value among software vendors. http://www.redhat.com/promo/vendor/
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
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.
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
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.
- -- Stephen Gallagher RHCE 804006346421761
Delivering value year after year. Red Hat ranks #1 in value among software vendors. http://www.redhat.com/promo/vendor/
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 12/02/2010 09:21 AM, Stephen Gallagher wrote:
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.
I take that back. I see it's supposed to be using $(srcdir)/po/LINGUAS, however something is broken, because 'make distcheck' fails with:
(cd src/man && make top_distdir=../../sssd-1.5.0 distdir=../../sssd-1.5.0/src/man \ am__remove_distdir=: am__skip_length_check=: am__skip_mode_fix=: distdir) /bin/sed: can't read ../../../src/man/po/LINGUAS: No such file or directory make[1]: Entering directory `/home/bos/sgallagh/workspace/sssd/x86_64/src/man' make[1]: *** No rule to make target `../../../src/man/po/sssd-docs.pot', needed by `distdir'. Stop.
- -- Stephen Gallagher RHCE 804006346421761
Delivering value year after year. Red Hat ranks #1 in value among software vendors. http://www.redhat.com/promo/vendor/
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 12/02/2010 03:25 PM, Stephen Gallagher wrote:
I take that back. I see it's supposed to be using $(srcdir)/po/LINGUAS, however something is broken, because 'make distcheck' fails with:
Sumit helped me fix this bug (srcdir/builddir mismatch). New patch attached.
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 12/14/2010 08:00 AM, Jakub Hrozek wrote:
On 12/02/2010 03:25 PM, Stephen Gallagher wrote:
I take that back. I see it's supposed to be using $(srcdir)/po/LINGUAS, however something is broken, because 'make distcheck' fails with:
Sumit helped me fix this bug (srcdir/builddir mismatch). New patch attached.
I think you sent the wrong patch. I'm still seeing the 'make distcheck failure'
Interdiff only says:
- --- b/contrib/sssd.spec.in +++ b/contrib/sssd.spec.in @@ -106,7 +106,6 @@ --disable-rpath
make %{?_smp_mflags} - -make translated-manpages
%check export CK_TIMEOUT_MULTIPLIER=10 @@ -120,6 +119,7 @@ --disable-rpath
make %{?_smp_mflags} +make translated-manpages
%check export CK_TIMEOUT_MULTIPLIER=10
- -- Stephen Gallagher RHCE 804006346421761
Delivering value year after year. Red Hat ranks #1 in value among software vendors. http://www.redhat.com/promo/vendor/
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
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.
Jakub
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 12/15/2010 08:15 AM, Jakub Hrozek wrote:
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.
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.
- -- Stephen Gallagher RHCE 804006346421761
Delivering value year after year. Red Hat ranks #1 in value among software vendors. http://www.redhat.com/promo/vendor/
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 12/15/2010 09:48 PM, Stephen Gallagher wrote:
On 12/15/2010 08:15 AM, Jakub Hrozek wrote:
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.
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 SIGNED MESSAGE----- Hash: SHA1
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.
- -- Stephen Gallagher RHCE 804006346421761
Delivering value year after year. Red Hat ranks #1 in value among software vendors. http://www.redhat.com/promo/vendor/
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
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 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.
- -- Stephen Gallagher RHCE 804006346421761
Delivering value year after year. Red Hat ranks #1 in value among software vendors. http://www.redhat.com/promo/vendor/
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/ -----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----- _______________________________________________ sssd-devel mailing list sssd-devel@lists.fedorahosted.org https://fedorahosted.org/mailman/listinfo/sssd-devel
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
Stephen Gallagher RHCE 804006346421761
Delivering value year after year. Red Hat ranks #1 in value among software vendors. http://www.redhat.com/promo/vendor/ -----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----- _______________________________________________ sssd-devel mailing list sssd-devel@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/ -----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----- _______________________________________________ sssd-devel mailing list sssd-devel@lists.fedorahosted.org https://fedorahosted.org/mailman/listinfo/sssd-devel
From 7be9f787dff5af9fe24218e093a840dce5e8407b Mon Sep 17 00:00:00 2001 From: Sumit Bose sbose@redhat.com Date: Mon, 20 Dec 2010 13:42:07 +0100 Subject: [PATCH] Update man pages pot file during make dist
src/man/Makefile.am | 5 +++-- 1 files changed, 3 insertions(+), 2 deletions(-)
diff --git a/src/man/Makefile.am b/src/man/Makefile.am index f66babe..2f2b295 100644 --- a/src/man/Makefile.am +++ b/src/man/Makefile.am @@ -60,7 +60,7 @@ PO4A_TRANSLATE_OPTS = --option doctype=docbook -f docbook
# The list of files to translate -XML_DOC=$(wildcard *.xml) $(wildcard include/*.xml) +XML_DOC=$(wildcard $(srcdir)/*.xml) $(wildcard $(srcdir)/include/*.xml) LANGUAGES=$(shell $(SED) 's/#.*//' $(LINGUAS))
EXTRA_DIST += \ @@ -125,7 +125,8 @@ translate: translate-xml.stamp done \ fi
-dist-hook: translate-xml.stamp +distdir-hook:
- $(MAKE) update-po
clean-local: clean-local-@USE_NLS@ distclean-local: clean-local-@USE_NLS@ -- 1.7.3.2
sssd-devel mailing list sssd-devel@lists.fedorahosted.org https://fedorahosted.org/mailman/listinfo/sssd-devel
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:
-----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
Sorry, found another issue. I've added the other two related patches for your convenience, too.
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/ -----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----- _______________________________________________ sssd-devel mailing list sssd-devel@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: >>>> 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
Sorry, found another issue. I've added the other two related patches for your convenience, too.
bye, Sumit
Sorry, this is still a Nack.
Patch 0001 doesn't apply cleanly on master. I had to tweak it to get it to work (conflict with the DISTCHECK_CONFIGURE_FLAGS in the patch made it fail to apply)
Once applied, I ran the following test:
cd /dev/shm rm -Rf sssd; mkdir sssd; cd sssd ~/workspace/sssd/configure && make distcheck
This failed with:
make[2]: Leaving directory `/dev/shm/sssd/sssd-1.5.0/_build/po' (cd src/man && make top_distdir=../../sssd-1.5.0 distdir=../../sssd-1.5.0/src/man \ am__remove_distdir=: am__skip_length_check=: am__skip_mode_fix=: distdir) make[2]: Entering directory `/dev/shm/sssd/sssd-1.5.0/_build/src/man' /bin/mkdir -p po for lang in cs; do \ /bin/mkdir -p $lang; \ /bin/mkdir -p $lang/include; \ done touch prepare.stamp rm -f po/po4a.cfg echo "[po4a_langs] cs" >> po/po4a.cfg echo "[po4a_paths] ../../../src/man/po/sssd-docs.pot $lang:../../../src/man/po/$lang.po" >> po/po4a.cfg for file in ../../../src/man/sss_groupmod.8.xml ../../../src/man/sssd.conf.5.xml ../../../src/man/sss_groupshow.8.xml ../../../src/man/sssd-ldap.5.xml ../../../src/man/pam_sss.8.xml ../../../src/man/sssd_krb5_locator_plugin.8.xml ../../../src/man/sssd-simple.5.xml ../../../src/man/sssd-ipa.5.xml ../../../src/man/sss_useradd.8.xml ../../../src/man/sss_obfuscate.8.xml ../../../src/man/sssd.8.xml ../../../src/man/sssd-krb5.5.xml ../../../src/man/sss_groupadd.8.xml ../../../src/man/sss_userdel.8.xml ../../../src/man/sss_groupdel.8.xml ../../../src/man/sss_usermod.8.xml ../../../src/man/include/service_discovery.xml ../../../src/man/include/upstream.xml ../../../src/man/include/failover.xml ../../../src/man/include/param_help.xml; do \ echo "[type:docbook] $file $lang:$lang/$file" >> po/po4a.cfg; \ done po4a --option doctype=docbook --package-name sssd-docs --package-version 1.5.0 --msgid-bugs-address sssd-devel@redhat.com --copyright-holder "Red Hat" --no-backups --no-translations ./po/po4a.cfg Can't move /tmp/vKC8899LFt to ../../../src/man/po/sssd-docs.pot: Permission denied. make[2]: *** [../../../src/man/po/sssd-docs.pot] Error 13 make[2]: Leaving directory `/dev/shm/sssd/sssd-1.5.0/_build/src/man' make[1]: *** [distdir] Error 1 make[1]: Leaving directory `/dev/shm/sssd/sssd-1.5.0/_build' make: *** [distcheck] Error 1
- -- Stephen Gallagher RHCE 804006346421761
Delivering value year after year. Red Hat ranks #1 in value among software vendors. http://www.redhat.com/promo/vendor/
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.
bye, Sumit
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.
bye, Sumit
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
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.
All of my other tests seemed to be working smoothly, so hopefully this is the last iteration.
- -- Stephen Gallagher RHCE 804006346421761
Delivering value year after year. Red Hat ranks #1 in value among software vendors. http://www.redhat.com/promo/vendor/
On Wed, Dec 22, 2010 at 11:52:01AM -0500, Stephen Gallagher wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
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.
Stephen Gallagher RHCE 804006346421761
Delivering value year after year. Red Hat ranks #1 in value among software vendors. http://www.redhat.com/promo/vendor/ -----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----- _______________________________________________ sssd-devel mailing list sssd-devel@lists.fedorahosted.org https://fedorahosted.org/mailman/listinfo/sssd-devel
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 12/22/2010 12:07 PM, Sumit Bose wrote:
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.
Everything looks good now. I hereby grant the long-awaited ack.
- -- Stephen Gallagher RHCE 804006346421761
Delivering value year after year. Red Hat ranks #1 in value among software vendors. http://www.redhat.com/promo/vendor/
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 12/22/2010 01:05 PM, Stephen Gallagher wrote:
Everything looks good now. I hereby grant the long-awaited ack.
Pushed to master.
- -- 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@lists.fedorahosted.org
-
Jakub Hrozek -
Stephen Gallagher -
Sumit Bose