Bug#682580: [bug-gettext] Bug#682580: xgettext: fails to properly replace some placeholders in output .pot (PACKAGE, YEAR, C. HOLDER) (fwd)

2015-09-14 Thread Francesco Poli 
On Thu, 03 Sep 2015 10:19:45 +0900 Daiki Ueno wrote:

> Francesco Poli  writes:
[...]
> > Please support these cases as well, at least by adding a
> > --same-copyright option (without arguments) that produces
> > a .pot file with
> >
> >   # Copyright (C) YEAR THE PACKAGE'S COPYRIGHT HOLDER
> >
> > replaced by
> >
> >   # This file is copyrighted by the PACKAGE package's copyright holders
> >   # and by the translators listed below.
> >
> > Of course, PACKAGE should be replaced by the argument of the
> > --package-name option, where available.
> 
> If I were going to address all such cases, I would rather add a single
> general-purpose --header-file=FILE option.  That would be equivalent to
> my proposal using sed, and I don't think it worth doing since the sed
> snippet is fairly simple.

I see that my bug report (on the Debian BTS) has been closed, since the
new version 0.19.6 incorporates the changes that you (Daiki)
implemented.

As I said, I am not especially happy with the result: xgettext has some
options to insert copyright notices into the generated .pot file, but
it only does half the job and needs "help" from external tools,
whenever there are multiple copyright holders.

Anyway, I give up: it seems that I failed to persuade you (Daiki) to
better support general use cases...   :-(

Bye.


-- 
 http://www.inventati.org/frx/
 There's not a second to spare! To the laboratory!
. Francesco Poli .
 GnuPG key fpr == CA01 1147 9CD2 EFDF FB82  3925 3E1C 27E1 1F69 BFFE


pgpznWhWGlMr3.pgp
Description: PGP signature

Bug#682580: [bug-gettext] Bug#682580: xgettext: fails to properly replace some placeholders in output .pot (PACKAGE, YEAR, C. HOLDER) (fwd)

2015-09-02 Thread Daiki Ueno 
Francesco Poli  writes:

>> I don't think it is boring nor error prone -- it just means to
>> concatenate two files.
>
> Well, not really just a file concatenation... More a job for sed or
> awk, as I said. But this would be the case, if you automated the task.

I actually pushed this to the git master:
http://git.savannah.gnu.org/cgit/gettext.git/commit/?id=e4ec2e69

It is only 5 lines and even smaller than a similar example quoted in the
Autoconf manual:
http://www.gnu.org/savannah-checkouts/gnu/autoconf/manual/autoconf-2.69/html_node/Installation-Directory-Variables.html#Installation-Directory-Variables

> Please support these cases as well, at least by adding a
> --same-copyright option (without arguments) that produces
> a .pot file with
>
>   # Copyright (C) YEAR THE PACKAGE'S COPYRIGHT HOLDER
>
> replaced by
>
>   # This file is copyrighted by the PACKAGE package's copyright holders
>   # and by the translators listed below.
>
> Of course, PACKAGE should be replaced by the argument of the
> --package-name option, where available.

If I were going to address all such cases, I would rather add a single
general-purpose --header-file=FILE option.  That would be equivalent to
my proposal using sed, and I don't think it worth doing since the sed
snippet is fairly simple.

Regards,
-- 
Daiki Ueno

Bug#682580: [bug-gettext] Bug#682580: xgettext: fails to properly replace some placeholders in output .pot (PACKAGE, YEAR, C. HOLDER) (fwd)

2015-09-02 Thread Francesco Poli 
On Mon, 31 Aug 2015 08:21:29 +0900 Daiki Ueno wrote:

> Francesco Poli  writes:
> 
> >   • I post-process the .pot template generated by xgettext and insert
> > the correct copyright notices: this can be done by hand (boring and
> > error prone!) or automatically, for instance, with some sed or awk
> > trick
> 
> I don't think it is boring nor error prone -- it just means to
> concatenate two files.

Well, not really just a file concatenation... More a job for sed or
awk, as I said. But this would be the case, if you automated the task.

I am convinced that doing the substitution *manually* is indeed boring
and error prone.

> As long as you use autotools, it would be
> automatically done by po/Makefile.in.in installed by autopoint.

You know, users of xgettext do not necessarily also use autotools...

> 
> So, my question was, why it is not sufficient for your use-case?  Is
> there any practical reason you have to call xgettext individually,
> e.g. your build system doesn't support post-processing for some reason?

I am not thinking about a build system with no post-processing support.
But what strikes me as weird is that xgettext has some support for
inserting appropriate copyright notices, but then leaves the job
half-done for all but very simple cases.

> 
> > At this point I am really puzzled. The --copyright-holder option is
> > currently inadequate to make sure the .pot template file has the
> > correct copyright notices in place.
> 
> The current behavior of --copyright-holder is useful for some use-cases,
> e.g., FSF-copyrighted packages (like coreutils, gettext), where
> translations are also attributed to the same copyright holder.  In that
> case, it is natural to leave the responsibility to rewrite YEAR to the
> translators.

But, as I said, there are many less simple cases out there.

Please support these cases as well, at least by adding a
--same-copyright option (without arguments) that produces
a .pot file with

  # Copyright (C) YEAR THE PACKAGE'S COPYRIGHT HOLDER

replaced by

  # This file is copyrighted by the PACKAGE package's copyright holders
  # and by the translators listed below.

Of course, PACKAGE should be replaced by the argument of the
--package-name option, where available.


Could at least this be done?

Thanks for your time and patience.


-- 
 http://www.inventati.org/frx/
 There's not a second to spare! To the laboratory!
. Francesco Poli .
 GnuPG key fpr == CA01 1147 9CD2 EFDF FB82  3925 3E1C 27E1 1F69 BFFE


pgpTEXGP14WcX.pgp
Description: PGP signature

Bug#682580: [bug-gettext] Bug#682580: xgettext: fails to properly replace some placeholders in output .pot (PACKAGE, YEAR, C. HOLDER) (fwd)

2015-08-30 Thread Francesco Poli 
On Thu, 27 Aug 2015 12:16:10 +0900 Daiki Ueno wrote:

 Francesco Poli invernom...@paranoici.org writes:
 
  I would definitely prefer to see xgettext changed, rather than having
  to do sed or awk tricks to post-process its output.
  So, to me, (4) is the least preferred option.
 
 Sorry, I am not convinced.  Again, I am reluctant to add a new option to
 xgettext for a niche feature, without good reason.  Do you have any
 practical reason for that?

I thought the practical reason was clear: I want to automatically
generate a .pot template file from the program source, ready to be sent
to prospective translators.
xgettext seems to be a suitable tool to achieve this goal.
But the .pot template has to include the correct copyright notices,
assuming it indeed has to include any copyright notice.

Just having a .pot template file which says

  # Copyright (C) YEAR Python Software Foundation

when there are actually more copyright holders, is misleading and
confusing (and is vague with respect to the copyright years).

So, I am left with two options:

  • I post-process the .pot template generated by xgettext and insert
the correct copyright notices: this can be done by hand (boring and
error prone!) or automatically, for instance, with some sed or awk
trick (but then, why does xgettext have the --copyright-holder option
in the first place?!?)

  • I tell prospective translators to edit the fake copyright notice
and replace it with the correct copyright notices (boring for
translators and even more error prone!)


At this point I am really puzzled. The --copyright-holder option is
currently inadequate to make sure the .pot template file has the
correct copyright notices in place.

Hence, xgettext should have appropriate options to insert the correct
copyright notices.

Or, alternatively, it should drop the --copyright-holder option
entirely and produce a .pot file with

  # Copyright (C) YEAR THE PACKAGE'S COPYRIGHT HOLDER

replaced by

  # This file is copyrighted by the PACKAGE package's copyright holders
  # and by the translators listed below.

Of course, PACKAGE should be replaced by the argument of the
--package-name option, where available.

This way, we could do the following:

  $ xgettext --package-name=myapplication --package-version=0.1 \
 --language=python myapplication.py -o myapplication3.pot
  $ cat myapplication3.pot 
  # SOME DESCRIPTIVE TITLE.
  # This file is copyrighted by the myapplication package's copyright holders
  # and by the translators listed below.
  # This file is distributed under the same license as the myapplication 
package.
  # FIRST AUTHOR EMAIL@ADDRESS, YEAR.
  #
  #, fuzzy
  msgid 
  msgstr 
  Project-Id-Version: myapplication 0.1\n
  Report-Msgid-Bugs-To: \n
  POT-Creation-Date: 2012-07-23 22:49+0200\n
  PO-Revision-Date: YEAR-MO-DA HO:MI+ZONE\n
  Last-Translator: FULL NAME EMAIL@ADDRESS\n
  Language-Team: LANGUAGE l...@li.org\n
  Language: \n
  MIME-Version: 1.0\n
  Content-Type: text/plain; charset=CHARSET\n
  Content-Transfer-Encoding: 8bit\n

  #: myapplication.py:6
  msgid This is a translatable string.
  msgstr 


 
 My reasons for preferring (4) to (3) are:
[...]
 - that fits better in the Unix philosophy, which suggests to accomplish
   larger tasks by combining small tools

As far as I understand the Unix design philosophy, there should be
small tools that do one thing (or, at most, a few things) and do it
well.
xgettext should generate a .pot template file and generate a good one.
In my own humble opinion, xgettext is currently failing at generating a
good .pot template file, as far as copyright notices are concerned.

I hope all this makes sense to you.
Please let me know, thanks for your time and patience!   :-)



-- 
 http://www.inventati.org/frx/
 There's not a second to spare! To the laboratory!
. Francesco Poli .
 GnuPG key fpr == CA01 1147 9CD2 EFDF FB82  3925 3E1C 27E1 1F69 BFFE


pgps2sfuByZqa.pgp
Description: PGP signature

Bug#682580: [bug-gettext] Bug#682580: xgettext: fails to properly replace some placeholders in output .pot (PACKAGE, YEAR, C. HOLDER) (fwd)

2015-08-30 Thread Daiki Ueno 
Francesco Poli invernom...@paranoici.org writes:

   • I post-process the .pot template generated by xgettext and insert
 the correct copyright notices: this can be done by hand (boring and
 error prone!) or automatically, for instance, with some sed or awk
 trick

I don't think it is boring nor error prone -- it just means to
concatenate two files.  As long as you use autotools, it would be
automatically done by po/Makefile.in.in installed by autopoint.

So, my question was, why it is not sufficient for your use-case?  Is
there any practical reason you have to call xgettext individually,
e.g. your build system doesn't support post-processing for some reason?

 At this point I am really puzzled. The --copyright-holder option is
 currently inadequate to make sure the .pot template file has the
 correct copyright notices in place.

The current behavior of --copyright-holder is useful for some use-cases,
e.g., FSF-copyrighted packages (like coreutils, gettext), where
translations are also attributed to the same copyright holder.  In that
case, it is natural to leave the responsibility to rewrite YEAR to the
translators.

Regards,
-- 
Daiki Ueno

Bug#682580: [bug-gettext] Bug#682580: xgettext: fails to properly replace some placeholders in output .pot (PACKAGE, YEAR, C. HOLDER) (fwd)

2015-08-26 Thread Francesco Poli 
On Tue, 25 Aug 2015 11:58:31 +0900 Daiki Ueno wrote:

 Francesco Poli invernom...@paranoici.org writes:
 
  I'd definitely vote for 3, but, wait!, even better, I would add a new
  option named --copyright-notices (or maybe there's a better name, I
  don't know...), where all the copyright notices may be specified with
  their own line breaks, as in:
 
 I'm afraid it would break configurability through po/Makevars, since a
 make variable including a multi-line argument cannot be used in a
 portable Makefile.

That's unfortunate.

 
 By the way, if we chose (4), the patch would be like the attached, which
 reads the file po/HEADER (if exists) and insert the content into POT
 file.  The file would look like:
 
   # Copyright (C) 2001-2012 Python Software Foundation.
   # Copyright (C) 2000 BeOpen.com.
   # Copyright (C) 1995-2000 Corporation for National Research Initiatives.
   # Copyright (C) 1991-1995 Stichting Mathematisch Centrum.
 
 If this is sufficient, I would revert the multiple --copyright-holder
 change, as I am reluctant to add new option to xgettext, which already
 has too many options.

I would definitely prefer to see xgettext changed, rather than having
to do sed or awk tricks to post-process its output.
So, to me, (4) is the least preferred option.

Since multi-line arguments cause problems, maybe an option named
--copyright-notice could be implemented in xgettext, to be used
multiple times as in:

  $ xgettext --copyright-notice=Copyright (C) 2001-2012 Python Software 
Foundation. \
 --copyright-notice=Copyright (C) 2000 BeOpen.com. \
 --copyright-notice=Copyright (C) 1995-2000 Corporation for 
National Research Initiatives. \
 --copyright-notice=Copyright (C) 1991-1995 Stichting Mathematisch 
Centrum. \
 --package-name=myapplication --package-version=0.1 \
 --language=python myapplication.py -o myapplication2.pot

Again, this would cause myapplication2.pot to include:

  # Copyright (C) 2001-2012 Python Software Foundation.
  # Copyright (C) 2000 BeOpen.com.
  # Copyright (C) 1995-2000 Corporation for National Research Initiatives.
  # Copyright (C) 1991-1995 Stichting Mathematisch Centrum.

in stead of:

  # Copyright (C) YEAR THE PACKAGE'S COPYRIGHT HOLDER


I hope you think this is a good suggestion.
Please let me know, thanks for your time!


-- 
 http://www.inventati.org/frx/
 There's not a second to spare! To the laboratory!
. Francesco Poli .
 GnuPG key fpr == CA01 1147 9CD2 EFDF FB82  3925 3E1C 27E1 1F69 BFFE


pgpSr4uEzrHcX.pgp
Description: PGP signature

Bug#682580: [bug-gettext] Bug#682580: xgettext: fails to properly replace some placeholders in output .pot (PACKAGE, YEAR, C. HOLDER) (fwd)

2015-08-26 Thread Daiki Ueno 
Francesco Poli invernom...@paranoici.org writes:

 I would definitely prefer to see xgettext changed, rather than having
 to do sed or awk tricks to post-process its output.
 So, to me, (4) is the least preferred option.

Sorry, I am not convinced.  Again, I am reluctant to add a new option to
xgettext for a niche feature, without good reason.  Do you have any
practical reason for that?

My reasons for preferring (4) to (3) are:

- we could give more control to users, for example, inserting arbitrary
  text other than copyright notice

- we are already using sed in po/Makefile.in.in, as in en@quot rules

- having multiple --copyright-foo options doesn't seem to play nicely
  with the feature supplying COPYRIGHT_HOLDER from autoconf (bug 764483)

- that fits better in the Unix philosophy, which suggests to accomplish
  larger tasks by combining small tools

Regards,
-- 
Daiki Ueno

Bug#682580: [bug-gettext] Bug#682580: xgettext: fails to properly replace some placeholders in output .pot (PACKAGE, YEAR, C. HOLDER) (fwd)

2015-08-24 Thread Daiki Ueno 
Francesco Poli invernom...@paranoici.org writes:

 Besides that, the patch seems to only address point (B) of my original
 bug report (multiple copyright holders), but not point (A) or point (C)
 (YEAR and PACKAGE placeholders). Or am I misreading the patch?

As (C) is trivial, I've pushed a fix:
http://git.savannah.gnu.org/cgit/gettext.git/commit/?id=3a08add5

However, I have not yet decided how to fix (A).  A few options I have in
mind are:

1. add --copyright-year option

2. extend --copyright-holder option so it can have a full copyright
  notice, if the argument is in the form Copyright (C) 

3. add a new option, say --add-copyright, which takes an argument in the
  form YEAR COPYRIGHT_HOLDER.

4. don't change xgettext, but modify po/Makefile.in.in to insert
   copyright comments, using sed

I think (1) is not very useful, particularly when multiple copyright
holders are supplied, and (2) is a bit heuristic.

So, I prefer (3) or (4), as both of them could allow users to specify it
through po/Makevars, though (3) is a bit too specific.

Suggestions would be appreciated.

Regards,
-- 
Daiki Ueno

Bug#682580: [bug-gettext] Bug#682580: xgettext: fails to properly replace some placeholders in output .pot (PACKAGE, YEAR, C. HOLDER) (fwd)

2015-08-24 Thread Francesco Poli 
On Mon, 24 Aug 2015 18:05:17 +0900 Daiki Ueno wrote:

 Francesco Poli invernom...@paranoici.org writes:
 
  Besides that, the patch seems to only address point (B) of my original
  bug report (multiple copyright holders), but not point (A) or point (C)
  (YEAR and PACKAGE placeholders). Or am I misreading the patch?
 
 As (C) is trivial, I've pushed a fix:
 http://git.savannah.gnu.org/cgit/gettext.git/commit/?id=3a08add5

Many thanks for doing so!  :-)

 
 However, I have not yet decided how to fix (A).  A few options I have in
 mind are:
[...]
 
 3. add a new option, say --add-copyright, which takes an argument in the
   form YEAR COPYRIGHT_HOLDER.
[...]
 
 Suggestions would be appreciated.

I'd definitely vote for 3, but, wait!, even better, I would add a new
option named --copyright-notices (or maybe there's a better name, I
don't know...), where all the copyright notices may be specified with
their own line breaks, as in:


  $ xgettext --copyright-notices=Copyright (C) 2001-2012 Python Software 
Foundation.
Copyright (C) 2000 BeOpen.com.
Copyright (C) 1995-2000 Corporation for National Research Initiatives.
Copyright (C) 1991-1995 Stichting Mathematisch Centrum. \
 --package-name=myapplication --package-version=0.1 \
 --language=python myapplication.py -o myapplication2.pot

This would cause myapplication2.pot to include:

  # Copyright (C) 2001-2012 Python Software Foundation.
  # Copyright (C) 2000 BeOpen.com.
  # Copyright (C) 1995-2000 Corporation for National Research Initiatives.
  # Copyright (C) 1991-1995 Stichting Mathematisch Centrum.

in stead of:

  # Copyright (C) YEAR THE PACKAGE'S COPYRIGHT HOLDER

In other words, the --copyright-notice option would allow the user to
specify the copyright notices literally; xgettext would just prepend
each copyright notice line with the #  comment symbol.


Does it sound like a good suggestion?

-- 
 http://www.inventati.org/frx/
 There's not a second to spare! To the laboratory!
. Francesco Poli .
 GnuPG key fpr == CA01 1147 9CD2 EFDF FB82  3925 3E1C 27E1 1F69 BFFE


pgp_hz_FFxA9x.pgp
Description: PGP signature

Bug#682580: [bug-gettext] Bug#682580: xgettext: fails to properly replace some placeholders in output .pot (PACKAGE, YEAR, C. HOLDER) (fwd)

2015-08-24 Thread Daiki Ueno 
Francesco Poli invernom...@paranoici.org writes:

 I'd definitely vote for 3, but, wait!, even better, I would add a new
 option named --copyright-notices (or maybe there's a better name, I
 don't know...), where all the copyright notices may be specified with
 their own line breaks, as in:

I'm afraid it would break configurability through po/Makevars, since a
make variable including a multi-line argument cannot be used in a
portable Makefile.

By the way, if we chose (4), the patch would be like the attached, which
reads the file po/HEADER (if exists) and insert the content into POT
file.  The file would look like:

  # Copyright (C) 2001-2012 Python Software Foundation.
  # Copyright (C) 2000 BeOpen.com.
  # Copyright (C) 1995-2000 Corporation for National Research Initiatives.
  # Copyright (C) 1991-1995 Stichting Mathematisch Centrum.

If this is sufficient, I would revert the multiple --copyright-holder
change, as I am reluctant to add new option to xgettext, which already
has too many options.

Regards,
-- 
Daiki Ueno
From c8f8099dca7976fa26d7ac671446144f1fe044c9 Mon Sep 17 00:00:00 2001
From: Daiki Ueno u...@gnu.org
Date: Tue, 25 Aug 2015 11:47:20 +0900
Subject: [PATCH] po: Insert HEADER file into POT file

* gettext-tools/po/Makefile.in.in ($(DOMAIN).pot-update): Add
insert-header.sed to prerequisites.  Insert the contents of the file
HEADER.
(mostlyclean): Remove insert-header.sed.
---
 gettext-runtime/po/ChangeLog  | 6 ++
 gettext-runtime/po/Makefile.in.in | 8 +++-
 2 files changed, 13 insertions(+), 1 deletion(-)

diff --git a/gettext-runtime/po/ChangeLog b/gettext-runtime/po/ChangeLog
index e5d8f0d..73ccaa2 100644
--- a/gettext-runtime/po/ChangeLog
+++ b/gettext-runtime/po/ChangeLog
@@ -1,3 +1,9 @@
+2015-08-25  Daiki Ueno  u...@gnu.org
+
+	* Makefile.in.in ($(DOMAIN).pot-update): Add insert-header.sed to
+	prerequisites.  Insert the contents of the file HEADER.
+	(mostlyclean): Remove insert-header.sed.
+
 2015-07-10  Daiki Ueno  u...@gnu.org
 
 	* gettext 0.19.5 released.
diff --git a/gettext-runtime/po/Makefile.in.in b/gettext-runtime/po/Makefile.in.in
index c36f1ad..f4267ab 100644
--- a/gettext-runtime/po/Makefile.in.in
+++ b/gettext-runtime/po/Makefile.in.in
@@ -150,7 +150,7 @@ stamp-po: $(srcdir)/$(DOMAIN).pot
 # The determination of whether the package xyz is a GNU one is based on the
 # heuristic whether some file in the top level directory mentions GNU xyz.
 # If GNU 'find' is available, we avoid grepping through monster files.
-$(DOMAIN).pot-update: $(POTFILES) $(srcdir)/POTFILES.in remove-potcdate.sed
+$(DOMAIN).pot-update: $(POTFILES) $(srcdir)/POTFILES.in remove-potcdate.sed insert-header.sed
 	package_gnu=$(PACKAGE_GNU); \
 	test -n $$package_gnu || { \
 	  if { if (LC_ALL=C find --version) 2/dev/null | grep GNU /dev/null; then \
@@ -195,6 +195,11 @@ $(DOMAIN).pot-update: $(POTFILES) $(srcdir)/POTFILES.in remove-potcdate.sed
 	;; \
 	esac
 	test ! -f $(DOMAIN).po || { \
+	  if test -f $(srcdir)/HEADER; then \
+	sed -f insert-header.sed  $(DOMAIN).po  $(DOMAIN).1po  \
+	rm -f $(DOMAIN).po  \
+	mv $(DOMAIN).1po $(DOMAIN).po; \
+	  fi; \
 	  if test -f $(srcdir)/$(DOMAIN).pot; then \
 	sed -f remove-potcdate.sed  $(srcdir)/$(DOMAIN).pot  $(DOMAIN).1po  \
 	sed -f remove-potcdate.sed  $(DOMAIN).po  $(DOMAIN).2po  \
@@ -365,6 +370,7 @@ info dvi ps pdf html tags TAGS ctags CTAGS ID:
 
 mostlyclean:
 	rm -f remove-potcdate.sed
+	rm -f insert-header.sed
 	rm -f stamp-poT
 	rm -f core core.* $(DOMAIN).po $(DOMAIN).1po $(DOMAIN).2po *.new.po
 	rm -fr *.o
-- 
2.4.3

Bug#682580: [bug-gettext] Bug#682580: xgettext: fails to properly replace some placeholders in output .pot (PACKAGE, YEAR, C. HOLDER) (fwd)

2015-08-21 Thread Santiago Vila 
On Fri, Aug 21, 2015 at 05:49:00PM +0900, Daiki Ueno wrote:
 What else was remaining to close this bug?

In Debian we usually close a bug when there is a new package available
in unstable fixing the bug. This is what I will do for the Debian
package but of course you might have different rules for the gettext
bugzilla.

Bug#682580: [bug-gettext] Bug#682580: xgettext: fails to properly replace some placeholders in output .pot (PACKAGE, YEAR, C. HOLDER) (fwd)

2015-08-21 Thread Francesco Poli 
On Fri, 21 Aug 2015 13:06:11 +0200 Santiago Vila wrote:

 On Fri, Aug 21, 2015 at 05:49:00PM +0900, Daiki Ueno wrote:
  What else was remaining to close this bug?
 
 In Debian we usually close a bug when there is a new package available
 in unstable fixing the bug. This is what I will do for the Debian
 package but of course you might have different rules for the gettext
 bugzilla.

Correct.

Besides that, the patch seems to only address point (B) of my original
bug report (multiple copyright holders), but not point (A) or point (C)
(YEAR and PACKAGE placeholders). Or am I misreading the patch?


-- 
 http://www.inventati.org/frx/
 There's not a second to spare! To the laboratory!
. Francesco Poli .
 GnuPG key fpr == CA01 1147 9CD2 EFDF FB82  3925 3E1C 27E1 1F69 BFFE


pgpBSgb69xWVI.pgp
Description: PGP signature

Bug#682580: [bug-gettext] Bug#682580: xgettext: fails to properly replace some placeholders in output .pot (PACKAGE, YEAR, C. HOLDER) (fwd)

2015-08-21 Thread Daiki Ueno 
Daiki Ueno u...@gnu.org writes:

 By the way, after looking into the history and the documentation more
 closely, I realized that my argument on copyright notice was pointless.
 I'm sorry.  I'm now in favor of adding support for multiple copyright
 holders, like the attached patch, which makes xgettext allow multiple
 --copyright-holder options.

Sorry for the delay.  I've landed the patch in upstream:
http://git.savannah.gnu.org/cgit/gettext.git/commit/?id=15363cca

What else was remaining to close this bug?

Regards,
-- 
Daiki Ueno

Bug#682580: [bug-gettext] Bug#682580: xgettext: fails to properly replace some placeholders in output .pot (PACKAGE, YEAR, C. HOLDER) (fwd)

2015-07-20 Thread Francesco Poli 
On Mon, 20 Jul 2015 08:10:25 +0900 Daiki Ueno wrote:

 Francesco Poli invernom...@paranoici.org writes:
 
[...]
  What I really fail to understand is: if some placeholders are replaced
  by
 
xgettext --copyright-holder=Python Software Foundation \
 --package-name=myapplication --package-version=0.1 \
 --language=python myapplication.py -o myapplication2.pot
 
  why other placeholders are not touched at all?!?
 
 First of all, I would repeat that xgettext prepares a template for
 translators, not a real content.  If you want to put the real content
 there, you can easily adjust it by using an Automake hook.
 
 With the above example, the replaced placeholders (i.e., COPYRIGHT
 HOLDER, PACKAGE, VERSION) are common for all translations derived from
 the package source code.
 
 On the other hand, xgettext intentionally doesn't provide a way to touch
 some placeholders (YEAR, FULL NAME, LANGUAGE, etc), because it shall be
 replaced by translators.  If you run:
 
   msginit -i myapplication2.pot -o ja.po
 
 then you will get ja.po with those placeholders properly replaced for
 translation.
 
 I think this is reasonable.  If you are still not satisfied with the
 answer, I'm afraid I doubt I could convince you of that.

Then I am afraid we will have to agree to disagree.

I am frankly having a hard time in seeing why the placeholder PACKAGE
should be replaced in one occurrence, but not in the other.
And in figuring out why the placeholder THE PACKAGE'S COPYRIGHT HOLDER
should be replaced, while the YEAR placeholder should be left untouched
in the *same* copyright notice.

Am I the only one who thinks that this is highly asymmetrical and
awkward?

I am honestly puzzled...



-- 
 http://www.inventati.org/frx/
 There's not a second to spare! To the laboratory!
. Francesco Poli .
 GnuPG key fpr == CA01 1147 9CD2 EFDF FB82  3925 3E1C 27E1 1F69 BFFE


pgpfwWgA88LdZ.pgp
Description: PGP signature

Bug#682580: [bug-gettext] Bug#682580: xgettext: fails to properly replace some placeholders in output .pot (PACKAGE, YEAR, C. HOLDER) (fwd)

2015-07-20 Thread Daiki Ueno 
Francesco Poli invernom...@paranoici.org writes:

 I am frankly having a hard time in seeing why the placeholder PACKAGE
 should be replaced in one occurrence, but not in the other.
 And in figuring out why the placeholder THE PACKAGE'S COPYRIGHT HOLDER
 should be replaced, while the YEAR placeholder should be left untouched
 in the *same* copyright notice.

I think it would be nice to fix them so those occurrences be replaced as
well.

By the way, after looking into the history and the documentation more
closely, I realized that my argument on copyright notice was pointless.
I'm sorry.  I'm now in favor of adding support for multiple copyright
holders, like the attached patch, which makes xgettext allow multiple
--copyright-holder options.

It is not straightforward to support multiple copyright holders in
po/Makevars, but it might be sufficient to supply it as part of
XGETTEXT_OPTIONS in combination with COPYRIGHT_HOLDER.

Regards,
-- 
Daiki Ueno
From 7aba166af10a4b64aa07d7f5afc91bdd239beef2 Mon Sep 17 00:00:00 2001
From: Daiki Ueno u...@gnu.org
Date: Tue, 21 Jul 2015 13:51:31 +0900
Subject: [PATCH] xgettext: Allow multiple copyright holders

* gettext-tools/src/xgettext.c (default_copyright_holder): New constant,
renamed from copyright_holder.
(copyright_holder): Define as a variable.
(main): Allow multiple --copyright-holder options.
(construct_header): Support multiple --copyright-holder options.
---
 gettext-tools/src/ChangeLog  |  8 +
 gettext-tools/src/xgettext.c | 70 
 2 files changed, 72 insertions(+), 6 deletions(-)

diff --git a/gettext-tools/src/ChangeLog b/gettext-tools/src/ChangeLog
index 6685505..74dd518 100644
--- a/gettext-tools/src/ChangeLog
+++ b/gettext-tools/src/ChangeLog
@@ -1,3 +1,11 @@
+2015-07-21  Daiki Ueno  u...@gnu.org
+
+	* xgettext.c (default_copyright_holder): New constant, renamed
+	from copyright_holder.
+	(copyright_holder): Define as a variable.
+	(main): Allow multiple --copyright-holder options.
+	(construct_header): Support multiple --copyright-holder options.
+
 2015-07-10  Daiki Ueno  u...@gnu.org
 
 	* gettext 0.19.5 released.
diff --git a/gettext-tools/src/xgettext.c b/gettext-tools/src/xgettext.c
index 9f5d300..8e303c7 100644
--- a/gettext-tools/src/xgettext.c
+++ b/gettext-tools/src/xgettext.c
@@ -124,7 +124,8 @@ message_list_ty *exclude;
 static int force_po;
 
 /* Copyright holder of the output file and the translations.  */
-static const char *copyright_holder = THE PACKAGE'S COPYRIGHT HOLDER;
+static const char *default_copyright_holder = THE PACKAGE'S COPYRIGHT HOLDER;
+static char *copyright_holder = NULL;
 
 /* Package name.  */
 static const char *package_name = NULL;
@@ -540,11 +541,19 @@ main (int argc, char *argv[])
 break;
 
   case CHAR_MAX + 1:/* --copyright-holder */
-copyright_holder = optarg;
+if (copyright_holder == NULL)
+  copyright_holder = xstrdup (optarg);
+else
+  {
+size_t total_len = strlen (copyright_holder) + 2 + strlen (optarg);
+copyright_holder = xrealloc (copyright_holder, total_len);
+strcat (copyright_holder, \n);
+strcat (copyright_holder, optarg);
+  }
 break;
 
   case CHAR_MAX + 2:/* --foreign-user */
-copyright_holder = ;
+copyright_holder = xstrdup ();
 break;
 
   case CHAR_MAX + 3:/* --from-code */
@@ -3559,13 +3568,61 @@ Content-Transfer-Encoding: 8bit\n,
 
   mp = message_alloc (NULL, , NULL, msgstr, strlen (msgstr) + 1, pos);
 
+  if (copyright_holder == NULL)
+copyright_holder = xstrdup (default_copyright_holder);
+
   if (copyright_holder[0] != '\0')
-comment = xasprintf (\
+{
+  size_t copyright_comment_len;
+  char *copyright_comment;
+  const char *p;
+  char *q;
+  size_t count = 1;
+
+  p = copyright_holder;
+  while (*p != '\0')
+{
+  p = strchr (p, '\n');
+  if (p == NULL)
+break;
+  count++;
+  p++;
+}
+
+  copyright_comment_len =
+strlen (copyright_holder) + strlen (Copyright (C) YEAR \n) * count;
+  copyright_comment = XNMALLOC (copyright_comment_len, char);
+
+  p = copyright_holder;
+  q = copyright_comment;
+  while (*p != '\0')
+{
+  char *newline = strchr (p, '\n');
+
+  q = stpcpy (q, Copyright (C) YEAR );
+  if (newline != NULL)
+{
+  *newline = '\0';
+  q = stpcpy (q, p);
+  q = stpcpy (q, \n);
+  p = newline + 1;
+}
+  else
+{
+  q = stpcpy (q, p);
+  q = stpcpy (q, \n);
+  break;
+}
+}
+
+  comment = xasprintf (\
 SOME DESCRIPTIVE TITLE.\n\
-Copyright (C) YEAR %s\n\
+%s\
 This file is distributed under the same license as the PACKAGE package.\n\
 FIRST AUTHOR EMAIL@ADDRESS,

Bug#682580: [bug-gettext] Bug#682580: xgettext: fails to properly replace some placeholders in output .pot (PACKAGE, YEAR, C. HOLDER) (fwd)

2015-07-19 Thread Daiki Ueno 
Francesco Poli invernom...@paranoici.org writes:

 dist-hook:
  echo # Copyright (C) ...  $(distdir)/po/hello.pot

 This would place the copyright notice at the end of the .pot file,
 wouldn't it? This is not necessarily what one would want to do.

I just gave you an example how it could be done.  It is easy to rewrite
it to put the copyright notice at the top, by using a temporary file.

 What I really fail to understand is: if some placeholders are replaced
 by

   xgettext --copyright-holder=Python Software Foundation \
--package-name=myapplication --package-version=0.1 \
--language=python myapplication.py -o myapplication2.pot

 why other placeholders are not touched at all?!?

First of all, I would repeat that xgettext prepares a template for
translators, not a real content.  If you want to put the real content
there, you can easily adjust it by using an Automake hook.

With the above example, the replaced placeholders (i.e., COPYRIGHT
HOLDER, PACKAGE, VERSION) are common for all translations derived from
the package source code.

On the other hand, xgettext intentionally doesn't provide a way to touch
some placeholders (YEAR, FULL NAME, LANGUAGE, etc), because it shall be
replaced by translators.  If you run:

  msginit -i myapplication2.pot -o ja.po

then you will get ja.po with those placeholders properly replaced for
translation.

I think this is reasonable.  If you are still not satisfied with the
answer, I'm afraid I doubt I could convince you of that.

Regards,
-- 
Daiki Ueno


-- 
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org

Bug#682580: [bug-gettext] Bug#682580: xgettext: fails to properly replace some placeholders in output .pot (PACKAGE, YEAR, C. HOLDER) (fwd)

2015-07-19 Thread Francesco Poli 
On Wed, 15 Jul 2015 07:03:51 +0900 Daiki Ueno wrote:

 Francesco Poli invernom...@paranoici.org writes:
 
  On Tue, 14 Jul 2015 10:48:20 +0900 Daiki Ueno wrote:
 
   I noticed that xgettext does not properly replace all the relevant
   placeholders in the output .pot file, when told to do so through
   command-line options.
  
  I doubt if it is meaningful to have a real copyright notice in an
  auto-generated template file.
 
  Well, the auto-generated .pot file is auto-generated by extracting many
  strings from a number of copyrighted source code files.
  I think that, in most cases, this implies that the .pot file is
  copyrighted and has the same copyright owners and copyright years as
  the source files.
  As a consequence, I think it makes definitely sense for it to carry
  real copyright notices (the same that cover the source files, in most
  cases).
 
 Yes.  However, xgettext is not intended for generating a real header
 entry, but for generating a template of a header entry.
 
 If one really wants to carry a real copyright notice in a POT file, she
 could easily do that from without xgettext, e.g., in the toplevel
 Makefile.am:
 
 dist-hook:
   echo # Copyright (C) ...  $(distdir)/po/hello.pot

I am not sure I understand this correctly.

This would place the copyright notice at the end of the .pot file,
wouldn't it? This is not necessarily what one would want to do.
Maybe one could use sed to replace the remaining placeholders, but
anyway it feels like a kludge...

What I really fail to understand is: if some placeholders are replaced
by

  xgettext --copyright-holder=Python Software Foundation \
   --package-name=myapplication --package-version=0.1 \
   --language=python myapplication.py -o myapplication2.pot

why other placeholders are not touched at all?!?



-- 
 http://www.inventati.org/frx/
 There's not a second to spare! To the laboratory!
. Francesco Poli .
 GnuPG key fpr == CA01 1147 9CD2 EFDF FB82  3925 3E1C 27E1 1F69 BFFE


pgpN6l4sLcEvQ.pgp
Description: PGP signature

Bug#682580: [bug-gettext] Bug#682580: xgettext: fails to properly replace some placeholders in output .pot (PACKAGE, YEAR, C. HOLDER) (fwd)

2015-07-14 Thread Daiki Ueno 
Francesco Poli invernom...@paranoici.org writes:

 On Tue, 14 Jul 2015 10:48:20 +0900 Daiki Ueno wrote:

  I noticed that xgettext does not properly replace all the relevant
  placeholders in the output .pot file, when told to do so through
  command-line options.
 
 I doubt if it is meaningful to have a real copyright notice in an
 auto-generated template file.

 Well, the auto-generated .pot file is auto-generated by extracting many
 strings from a number of copyrighted source code files.
 I think that, in most cases, this implies that the .pot file is
 copyrighted and has the same copyright owners and copyright years as
 the source files.
 As a consequence, I think it makes definitely sense for it to carry
 real copyright notices (the same that cover the source files, in most
 cases).

Yes.  However, xgettext is not intended for generating a real header
entry, but for generating a template of a header entry.

If one really wants to carry a real copyright notice in a POT file, she
could easily do that from without xgettext, e.g., in the toplevel
Makefile.am:

dist-hook:
echo # Copyright (C) ...  $(distdir)/po/hello.pot

Regards,
-- 
Daiki Ueno


-- 
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org

Bug#682580: [bug-gettext] Bug#682580: xgettext: fails to properly replace some placeholders in output .pot (PACKAGE, YEAR, C. HOLDER) (fwd)

2015-07-14 Thread Francesco Poli 
On Tue, 14 Jul 2015 10:48:20 +0900 Daiki Ueno wrote:

  I noticed that xgettext does not properly replace all the relevant
  placeholders in the output .pot file, when told to do so through
  command-line options.
 
 I doubt if it is meaningful to have a real copyright notice in an
 auto-generated template file.

Well, the auto-generated .pot file is auto-generated by extracting many
strings from a number of copyrighted source code files.
I think that, in most cases, this implies that the .pot file is
copyrighted and has the same copyright owners and copyright years as
the source files.
As a consequence, I think it makes definitely sense for it to carry
real copyright notices (the same that cover the source files, in most
cases).

 
 The gettext manual suggests that those placeholders shall be replaced by
 translators:
 http://www.gnu.org/software/gettext/manual/html_node/Header-Entry.html#Header-Entry

I think that the translation .po files are generally to be considered
as derivative works based on the .pot template file.
They are therefore copyrighted by the copyright owners of the .pot
template file *and* by the translator(s).

Hence, I think it makes sense that the translator(s) may *add* their own
copyright notices, if they feel like doing so, but they should _not_
*replace* the original copyright notices with their own notices...


Please let me know what you think about my reasoning.
Thanks for your time!


-- 
 http://www.inventati.org/frx/
 There's not a second to spare! To the laboratory!
. Francesco Poli .
 GnuPG key fpr == CA01 1147 9CD2 EFDF FB82  3925 3E1C 27E1 1F69 BFFE


pgp5NinsgRSIo.pgp
Description: PGP signature

Bug#682580: xgettext: fails to properly replace some placeholders in output .pot (PACKAGE, YEAR, C. HOLDER)

2012-07-23 Thread Francesco Poli (wintermute) 
Package: gettext
Version: 0.18.1.1-9
Severity: normal

Hello!
Thanks for maintaining gettext in Debian!


I noticed that xgettext does not properly replace all the relevant
placeholders in the output .pot file, when told to do so through
command-line options.

I'll try to explain the issues with a minimal test case.

Let's consider the example in the official Python library reference [1].
This should be minimal enough, I think.

[1] http://docs.python.org/library/gettext.html#module-gettext

Here's the example code:

  $ cat myapplication.py 
  import gettext
  gettext.bindtextdomain('myapplication', '/path/to/my/language/directory')
  gettext.textdomain('myapplication')
  _ = gettext.gettext
  # ...
  print _('This is a translatable string.')

Let's generate the corresponding .pot file:

  $ xgettext --language=python myapplication.py -o myapplication.pot
  $ cat myapplication.pot 
  # SOME DESCRIPTIVE TITLE.
  # Copyright (C) YEAR THE PACKAGE'S COPYRIGHT HOLDER
  # This file is distributed under the same license as the PACKAGE package.
  # FIRST AUTHOR EMAIL@ADDRESS, YEAR.
  #
  #, fuzzy
  msgid 
  msgstr 
  Project-Id-Version: PACKAGE VERSION\n
  Report-Msgid-Bugs-To: \n
  POT-Creation-Date: 2012-07-23 22:49+0200\n
  PO-Revision-Date: YEAR-MO-DA HO:MI+ZONE\n
  Last-Translator: FULL NAME EMAIL@ADDRESS\n
  Language-Team: LANGUAGE l...@li.org\n
  Language: \n
  MIME-Version: 1.0\n
  Content-Type: text/plain; charset=CHARSET\n
  Content-Transfer-Encoding: 8bit\n

  #: myapplication.py:6
  msgid This is a translatable string.
  msgstr 

This seems to be OK, but some placeholders should be replaced by actual
values.
It is my understanding that (before the .pot file is made available to
prospective translators who wish to start a localization in a new
language) the second line should be replaced by appropriate copyright
notices, and that PACKAGE and VERSION should be replaced by the actual
package name and version number.
Let's try with the following command-line options:

  $ xgettext --copyright-holder=Python Software Foundation \
 --package-name=myapplication --package-version=0.1 \
 --language=python myapplication.py -o myapplication2.pot
  $ cat myapplication2.pot 
  # SOME DESCRIPTIVE TITLE.
  # Copyright (C) YEAR Python Software Foundation
  # This file is distributed under the same license as the PACKAGE package.
  # FIRST AUTHOR EMAIL@ADDRESS, YEAR.
  #
  #, fuzzy
  msgid 
  msgstr 
  Project-Id-Version: myapplication 0.1\n
  Report-Msgid-Bugs-To: \n
  POT-Creation-Date: 2012-07-23 22:49+0200\n
  PO-Revision-Date: YEAR-MO-DA HO:MI+ZONE\n
  Last-Translator: FULL NAME EMAIL@ADDRESS\n
  Language-Team: LANGUAGE l...@li.org\n
  Language: \n
  MIME-Version: 1.0\n
  Content-Type: text/plain; charset=CHARSET\n
  Content-Transfer-Encoding: 8bit\n

  #: myapplication.py:6
  msgid This is a translatable string.
  msgstr 
  $ diff myapplication.pot myapplication2.pot 
  2c2
   # Copyright (C) YEAR THE PACKAGE'S COPYRIGHT HOLDER
  ---
   # Copyright (C) YEAR Python Software Foundation
  9c9
   Project-Id-Version: PACKAGE VERSION\n
  ---
   Project-Id-Version: myapplication 0.1\n

That's better, but there are a few issues left:

 (A) first of all, the YEAR placeholder in the second line is not
 replaced, and I failed to find an option to do so

 (B) secondly, I cannot find a way to include multi-line copyright
 notices, as in [2]:

 # Copyright (C) 2001-2012 Python Software Foundation.
 # Copyright (C) 2000 BeOpen.com.
 # Copyright (C) 1995-2000 Corporation for National Research Initiatives.
 # Copyright (C) 1991-1995 Stichting Mathematisch Centrum.

[2] http://docs.python.org/copyright.html

 (C) finally, the PACKAGE placeholder is correctly replaced in the
 Project-Id-Version field, but left untouched in the third line of
 myapplication2.pot


Please fix these issues and/or forward this bug report upstream,
as appropriate.

Thanks for your time!



-- System Information:
Debian Release: wheezy/sid
  APT prefers testing
  APT policy: (800, 'testing'), (500, 'unstable')
Architecture: amd64 (x86_64)

Kernel: Linux 3.2.0-3-amd64 (SMP w/2 CPU cores)
Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/bash

Versions of packages gettext depends on:
ii  dpkg1.16.4.3
ii  gettext-base0.18.1.1-9
ii  install-info4.13a.dfsg.1-10
ii  libasprintf0c2  0.18.1.1-9
ii  libc6   2.13-33
ii  libcroco3   0.6.5-1
ii  libgettextpo0   0.18.1.1-9
ii  libglib2.0-02.32.3-1
ii  libgomp14.7.1-2
ii  libncurses5 5.9-10
ii  libtinfo5   5.9-10
ii  libunistring0   0.9.3-5
ii  libxml2 2.8.0+dfsg1-4

Versions of packages gettext recommends:
ii  autopoint  0.18.1.1-9
ii  curl   7.26.0-1
ii  wget   1.13.4-3

Versions of packages gettext suggests:
pn  gettext-doc  none

-- no debconf information


-- 
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of