Re: fortran/5558: Error in POT file

Re: fortran/5558: Error in POT file

[Martin v. Loewis]

The following reply was made to PR fortran/5558; it has been noted by GNATS.

From: martin@v.loewis.de (Martin v. Loewis)
To: Zack Weinberg <zack@codesourcery.com>
Cc: gcc-gnats@gcc.gnu.org
Subject: Re: fortran/5558: Error in POT file
Date: 01 Feb 2002 10:20:43 +0100

 Zack Weinberg <zack@codesourcery.com> writes:
 
 > > I did not say they aren't substitution escapes, I said they are not
 > > meant for printf. Being substitution escapes is irrelevant for
 > > gettext.
 > 
 > However, they need to not get taken out just as much as printf escapes
 > do.
 
 That might well be, so there was a bug in this French
 translation. This bug is not the issue of this bug report, though.
 Michel Robitaille will mark all questionable messages as fuzzy, for
 later consideration. We should solve this problem now, before more
 translators run into the problem.
 
 > > Please consider integrating my proposed changes.
 > 
 > I tried them.  They don't do anything at all.  The generated .pot file
 > is identical with and without.
 
 What xgettext version are you using? With xgettext 0.10.40, on Linux,
 using the patch
 
 Index: lex.c
 ===================================================================
 RCS file: /cvs/gcc/egcs/gcc/f/lex.c,v
 retrieving revision 1.31
 diff -u -r1.31 lex.c
 --- lex.c	2002/01/02 23:18:57	1.31
 +++ lex.c	2002/02/01 08:34:18
 @@ -424,6 +424,7 @@
  		   && ((1 << (TYPE_PRECISION (integer_type_node) - (count - 1) * 4))
  		       <= (int) firstdig)))
  	{
 +	  /* xgettext:no-c-format */
  	  ffebad_start_msg_lex ("Hex escape at %0 out of range",
  				FFEBAD_severityPEDANTIC);
  	  ffelex_bad_here_ (0, line, column);
 @@ -457,6 +458,7 @@
        && TYPE_PRECISION (char_type_node) < HOST_BITS_PER_INT
        && code >= (1 << TYPE_PRECISION (char_type_node)))
      {
 +      /* xgettext:no-c-format */
        ffebad_start_msg_lex ("Escape sequence at %0 out of range for character",
  			    FFEBAD_severityFATAL);
        ffelex_bad_here_ (0, line, column);
 
 I get a template containing
 
 #: f/lex.c:428
 #, no-c-format
 msgid "Hex escape at %0 out of range"
 msgstr ""
 
 #: f/lex.c:462
 #, no-c-format
 msgid "Escape sequence at %0 out of range for character"
 msgstr ""
 
 Notice how c-format has been replaced with no-c-format.
 
 > I was not aware that this was the case - and the Turkish translator
 > had a similar problem, reported that removing the c-format marker
 > solved it, and has had no trouble uploading.
 
 In those cases (e.g. config/m88k/m88k.c:3063), msgfmt accepts the
 translation even with c-format (%C is taken as the wchar_t
 format). Therefore, the translator did not need to remove the c-format
 markers, and the robot accepted the translation since msgfmt found no
 errors even after putting them back in.
 
 Regards,
 Martin

Re: fortran/5558: Error in POT file

[pthomas]

Synopsis: Error in POT file

State-Changed-From-To: feedback->closed
State-Changed-By: pthomas
State-Changed-When: Mon Feb  4 15:29:44 2002
State-Changed-Why:
    All messages that might be mistaken to be printf style
    have been marked by the special comment.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=5558

Re: fortran/5558: Error in POT file

[Martin v. Loewis]

The following reply was made to PR fortran/5558; it has been noted by GNATS.

From: martin@v.loewis.de (Martin v. Loewis)
To: Toon Moene <toon@moene.indiv.nluug.nl>
Cc: Philipp Thomas <pthomas@suse.de>, toon@gcc.gnu.org, gcc-bugs@gcc.gnu.org,
   gcc-prs@gcc.gnu.org, nobody@gcc.gnu.org, gcc-gnats@gcc.gnu.org
Subject: Re: fortran/5558: Error in POT file
Date: 02 Feb 2002 01:13:58 +0100

 Toon Moene <toon@moene.indiv.nluug.nl> writes:
 
 > However, my question was:  Is this something that should be repaired in
 > gcc.pot, or is it something that should be repaired in the program that
 > generates gcc.pot.
 
 Neither, nor: it (insertion of incorrec c-format markup) should be
 corrected in the input that is used to generate gcc.pot.
 
 Regards,
 Martin

Re: fortran/5558: Error in POT file

[Philipp Thomas]

The following reply was made to PR fortran/5558; it has been noted by GNATS.

From: Philipp Thomas <pthomas@suse.de>
To: Toon Moene <toon@moene.indiv.nluug.nl>
Cc: toon@gcc.gnu.org, gcc-bugs@gcc.gnu.org, gcc-gnats@gcc.gnu.org
Subject: Re: fortran/5558: Error in POT file
Date: Fri, 1 Feb 2002 23:55:35 +0100

 * Toon Moene (toon@moene.indiv.nluug.nl) [20020201 23:50]:
 
 > However, my question was:  Is this something that should be repaired in
 > gcc.pot, or is it something that should be repaired in the program that
 > generates gcc.pot.
 
 Neither ;-) The respective lines in the *sources* have to be marked with
 that comment :( See the patch I sent on how it has to be done.
 
 Philipp
 
 -- 
 Philipp Thomas <pthomas@suse.de>
 Development, SuSE GmbH, Deutscherrnstr. 15-19, D-90429 Nuremberg, Germany
 
 Penguins shall save the dinosaurs
                           -- Handelsblatt about Linux on S/390

Re: fortran/5558: Error in POT file

[Toon Moene]

The following reply was made to PR fortran/5558; it has been noted by GNATS.

From: Toon Moene <toon@moene.indiv.nluug.nl>
To: Philipp Thomas <pthomas@suse.de>
Cc: toon@gcc.gnu.org, gcc-bugs@gcc.gnu.org, gcc-prs@gcc.gnu.org,
 	martin@v.loewis.de, nobody@gcc.gnu.org, gcc-gnats@gcc.gnu.org
Subject: Re: fortran/5558: Error in POT file
Date: Fri, 01 Feb 2002 23:51:00 +0100

 Philipp Thomas wrote:
 
 > * Toon Moene (toon@moene.indiv.nluug.nl) [20020201 22:20]:
 > 
 > > The problem is caused by the fact that several "proto" error messages in
 > > the Fortran front-end have "printf"-like "%" sequences.
 > >
 > > However, they definitely are *not* printf format strings.  Hence
 > > classifying them as c-format is wrong.
 
 > The solution is very easy. Before each of those lines put the comment
 > 
 > /* xgettext:no-c-format */
 > 
 > That will prevent xgettext ever mistakenly mark such a string as
 > printf-like. We could limit the marking to only those strings that xgettext
 > wrongly marked, but I'd say fix it once and for all.
 > 
 > Now who's going to do the dirty work, me or you? ;-)
 
 Well, yes, this I understood already :-)
 
 However, my question was:  Is this something that should be repaired in
 gcc.pot, or is it something that should be repaired in the program that
 generates gcc.pot.
 
 Just to prove this point :-) I changed one of the error messages that's
 listed in gcc.pot in the original program that contains this error
 message.
 
 -- 
 Toon Moene - mailto:toon@moene.indiv.nluug.nl - phoneto: +31 346 214290
 Saturnushof 14, 3738 XG  Maartensdijk, The Netherlands
 Maintainer, GNU Fortran 77: http://gcc.gnu.org/onlinedocs/g77_news.html
 Join GNU Fortran 95: http://g95.sourceforge.net/ (under construction)

Re: fortran/5558: Error in POT file

[Philipp Thomas]

The following reply was made to PR fortran/5558; it has been noted by GNATS.

From: Philipp Thomas <pthomas@suse.de>
To: Toon Moene <toon@moene.indiv.nluug.nl>
Cc: toon@gcc.gnu.org, gcc-bugs@gcc.gnu.org, gcc-prs@gcc.gnu.org,
	martin@v.loewis.de, nobody@gcc.gnu.org, gcc-gnats@gcc.gnu.org
Subject: Re: fortran/5558: Error in POT file
Date: Fri, 1 Feb 2002 22:49:31 +0100

 * Toon Moene (toon@moene.indiv.nluug.nl) [20020201 22:20]:
 
 > The problem is caused by the fact that several "proto" error messages in
 > the Fortran front-end have "printf"-like "%" sequences.
 > 
 > However, they definitely are *not* printf format strings.  Hence
 > classifying them as c-format is wrong.
 
 The solution is very easy. Before each of those lines put the comment
 
 /* xgettext:no-c-format */
 
 That will prevent xgettext ever mistakenly mark such a string as
 printf-like. We could limit the marking to only those strings that xgettext
 wrongly marked, but I'd say fix it once and for all.
 
 Now who's going to do the dirty work, me or you? ;-)
 
 Philipp
 
 -- 
 Philipp Thomas <pthomas@suse.de>
 Development, SuSE Linux AG, Deutscherrnstr. 15-19, D-90429 Nuremberg, Germany

Re: fortran/5558: Error in POT file

[Toon Moene]

The following reply was made to PR fortran/5558; it has been noted by GNATS.

From: Toon Moene <toon@moene.indiv.nluug.nl>
To: Philipp Thomas <pthomas@suse.de>
Cc: toon@gcc.gnu.org, gcc-bugs@gcc.gnu.org, gcc-prs@gcc.gnu.org,
 	martin@v.loewis.de, nobody@gcc.gnu.org, gcc-gnats@gcc.gnu.org
Subject: Re: fortran/5558: Error in POT file
Date: Fri, 01 Feb 2002 22:14:20 +0100

 Philipp Thomas wrote:
 
 > * toon@gcc.gnu.org (toon@gcc.gnu.org) [20020201 15:09]:
 > 
 > >     Philipp, can you deal with this bug report - thanks,
 > 
 > I'll take a look ASAP.
 
 The problem is caused by the fact that several "proto" error messages in
 the Fortran front-end have "printf"-like "%" sequences.
 
 However, they definitely are *not* printf format strings.  Hence
 classifying them as c-format is wrong.
 
 Here is the list (I also list the ones that are *not* given the
 qualifier "c-format", just to be sure):
 
 #: f/com.c:3116
 msgid "ASSIGN'ed label cannot fit into `%A' at %0 -- using wider
 sibling"
 msgstr ""
 
 #: f/com.c:15376
 msgid "%A from %B at %0%C"
 msgstr ""
 
 #: f/com.c:15695
 msgid "At %0, INCLUDE file %A exists, but is not readable"
 msgstr ""
 
 #: f/com.c:15729
 msgid "At %0, INCLUDE nesting too deep"
 msgstr ""
 
 #: f/expr.c:9611
 msgid "Two arithmetic operators in a row at %0 and %1 -- use
 parentheses"
 msgstr ""
 
 #: f/expr.c:9660
 msgid "Operator at %0 has lower precedence than that at %1 -- use
 parentheses"
 msgstr ""
 
 #: f/expr.c:10520
 msgid ""
 "Use .EQV./.NEQV. instead of .EQ./.NE. at %0 for LOGICAL operands at %1
 and %2"
 msgstr ""
 
 #: f/expr.c:10890
 msgid "Unsupported operand for ** at %1 -- converting to default
 INTEGER"
 msgstr ""
 
 #: f/implic.c:202
 msgid "Implicit declaration of `%A' at %0"
 msgstr ""
 
 #: f/lex.c:272
 msgid "The meaning of `\\x' (at %0) varies with -traditional"
 msgstr ""
 
 #: f/lex.c:322
 msgid "The meaning of `\\a' (at %0) varies with -traditional"
 msgstr ""
 
 #: f/lex.c:351
 msgid "Non-ISO-C-standard escape sequence `\\%A' at %0"
 msgstr ""
 
 #: f/lex.c:369
 msgid "Unknown escape sequence `\\%A' at %0"
 msgstr ""
 
 #: f/lex.c:377
 #, c-format
 msgid "Unterminated escape sequence `\\' at %0"
 msgstr ""
 
 #: f/lex.c:387
 msgid "Unknown escape sequence `\\' followed by char code 0x%A at %0"
 msgstr ""
 
 #: f/lex.c:414
 msgid "\\x used at %0 with no following hex digits"
 msgstr ""
 
 #: f/lex.c:427
 #, c-format
 msgid "Hex escape at %0 out of range"
 msgstr ""
 
 #: f/lex.c:460
 #, c-format
 msgid "Escape sequence at %0 out of range for character"
 msgstr ""
 
 #: f/lex.c:1465
 #, c-format
 msgid "Null character at %0 -- line ignored"
 msgstr ""
 
 #: f/stb.c:12077
 #, c-format
 msgid "INCLUDE at %0 not the only statement on the source line"
 msgstr ""
 
 #. ~~~Someday handle CHARACTER*1, CHARACTER*N
 #: f/ste.c:2635
 msgid "SELECT CASE on CHARACTER type (at %0) not supported -- sorry"
 msgstr ""
 
 #: f/stu.c:304
 msgid "Local adjustable symbol `%A' at %0"
 msgstr ""
 
 and all messages in f/bad.def.
 
 Hope this helps,
 
 -- 
 Toon Moene - mailto:toon@moene.indiv.nluug.nl - phoneto: +31 346 214290
 Saturnushof 14, 3738 XG  Maartensdijk, The Netherlands
 Maintainer, GNU Fortran 77: http://gcc.gnu.org/onlinedocs/g77_news.html
 Join GNU Fortran 95: http://g95.sourceforge.net/ (under construction)

Re: fortran/5558: Error in POT file

[Philipp Thomas]

The following reply was made to PR fortran/5558; it has been noted by GNATS.

From: Philipp Thomas <pthomas@suse.de>
To: toon@gcc.gnu.org, gcc-bugs@gcc.gnu.org, gcc-prs@gcc.gnu.org,
	martin@v.loewis.de, nobody@gcc.gnu.org, gcc-gnats@gcc.gnu.org
Cc:  
Subject: Re: fortran/5558: Error in POT file
Date: Fri, 1 Feb 2002 15:13:55 +0100

 * toon@gcc.gnu.org (toon@gcc.gnu.org) [20020201 15:09]:
 
 >     Philipp, can you deal with this bug report - thanks,
 
 I'll take a look ASAP.
 
 Philipp

Re: fortran/5558: Error in POT file

[toon]

Synopsis: Error in POT file

State-Changed-From-To: open->feedback
State-Changed-By: toon
State-Changed-When: Fri Feb  1 06:09:29 2002
State-Changed-Why:
    Not sure the error is in the indicated file - isn't this a
    generated file and the error in the generating program ?
    
    Philipp, can you deal with this bug report - thanks,
    
    Toon.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=5558

Re: fortran/5558: Error in POT file

[Zack Weinberg]

The following reply was made to PR fortran/5558; it has been noted by GNATS.

From: Zack Weinberg <zack@codesourcery.com>
To: "Martin v. Loewis" <martin@v.loewis.de>
Cc: gcc-gnats@gcc.gnu.org
Subject: Re: fortran/5558: Error in POT file
Date: Thu, 31 Jan 2002 16:59:44 -0800

 On Fri, Feb 01, 2002 at 01:30:34AM +0100, Martin v. Loewis wrote:
 > Zack Weinberg <zack@codesourcery.com> writes:
 > 
 > > You are mistaken.  Those %0's most definitely are substitution
 > > escapes.  They are not the same as printf substitutions, but they
 > > are substitutions nonetheless.
 > 
 > I did not say they aren't substitution escapes, I said they are not
 > meant for printf. Being substitution escapes is irrelevant for
 > gettext.
 
 However, they need to not get taken out just as much as printf escapes
 do.
 
 > > Having said that, I can see the problem - it thinks the escape is 
 > > "%0 o" instead of just "%0".  Your fix is to remove the "#, c-format"
 > > annotation.
 > 
 > Please consider integrating my proposed changes.
 
 I tried them.  They don't do anything at all.  The generated .pot file
 is identical with and without.
 
 > > I am attempting to get a more flexible mechanism for format-string
 > > handling put into gettext, but Bruno so far refuses to even consider
 > > it.
 > 
 > Please don't use the gcc catalog as a weapon in your fight with
 > Bruno. 
 
 ? Where did I say I was having a fight with Bruno?
 
 > (the robot will reject submissions if translators have removed
 > c-format markers; this was necessary to avoid security risks
 > resulting from buffer overruns).
 
 I was not aware that this was the case - and the Turkish translator
 had a similar problem, reported that removing the c-format marker
 solved it, and has had no trouble uploading.
 
 zw

Re: fortran/5558: Error in POT file

[Martin v. Loewis]

The following reply was made to PR fortran/5558; it has been noted by GNATS.

From: martin@v.loewis.de (Martin v. Loewis)
To: Zack Weinberg <zack@codesourcery.com>
Cc: gcc-gnats@gcc.gnu.org
Subject: Re: fortran/5558: Error in POT file
Date: 01 Feb 2002 01:30:34 +0100

 Zack Weinberg <zack@codesourcery.com> writes:
 
 > You are mistaken.  Those %0's most definitely are substitution
 > escapes.  They are not the same as printf substitutions, but they
 > are substitutions nonetheless.
 
 I did not say they aren't substitution escapes, I said they are not
 meant for printf. Being substitution escapes is irrelevant for
 gettext.
 
 > Having said that, I can see the problem - it thinks the escape is 
 > "%0 o" instead of just "%0".  Your fix is to remove the "#, c-format"
 > annotation.
 
 Please consider integrating my proposed changes. It is a plain bug in
 gcc that the provided catalog includes the c-format markup. This
 c-format is put in by your invocation of xgettext, so I would argue
 that it is your job to remove it. xgettext provides the mechanism to
 do this automatically (instead of doing it manually after each
 extraction), so please use them.
 
 > The second translation needs to have "%0" in it somewhere analogous to
 > where it appears in the English, or the error messages will not make
 > sense in context.
 
 Well, that would be then an error in the translation. There are many
 ways to deal with those; perhaps the best way (given the current
 tools) is to put a comment for the translators into the catalog that
 the %0 escape must occur in the translated message. If you want to
 create a custom to to perform this check automatically, that would be
 fine as well.
 
 > I am attempting to get a more flexible mechanism for format-string
 > handling put into gettext, but Bruno so far refuses to even consider
 > it.
 
 Please don't use the gcc catalog as a weapon in your fight with
 Bruno. Instead, please incorporate the requested changes, so that
 translators have a chance to submit their translations (the robot will
 reject submissions if translators have removed c-format markers; this
 was necessary to avoid security risks resulting from buffer overruns).
 
 Regards,
 Martin
 

Re: fortran/5558: Error in POT file

[Zack Weinberg]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1608 bytes --]

The following reply was made to PR fortran/5558; it has been noted by GNATS.

From: Zack Weinberg <zack@codesourcery.com>
To: martin@v.loewis.de
Cc: gcc-gnats@gcc.gnu.org
Subject: Re: fortran/5558: Error in POT file
Date: Thu, 31 Jan 2002 15:50:45 -0800

 On Thu, Jan 31, 2002 at 09:02:40PM -0000, martin@v.loewis.de wrote:
 
 > There are atleast two messages in the gcc catalog which are marked
 > as c-format, but are not intended to be used with printf. As a
 > result, msgfmt rejects the French translation. The specific examples
 > are
 > 
 > #: f/lex.c:427
 > #, c-format
 > msgid "Hex escape at %0 out of range"
 > msgstr "échappement hexadécimal à %0 hors gamme"
 > 
 > #: f/lex.c:460
 > #, fuzzy, c-format
 > msgid "Escape sequence at %0 out of range for character"
 > msgstr "séquence d'échappement hors gamme pour le type caractère"
 
 You are mistaken.  Those %0's most definitely are substitution
 escapes.  They are not the same as printf substitutions, but they are
 substitutions nonetheless.
 
 Having said that, I can see the problem - it thinks the escape is 
 "%0 o" instead of just "%0".  Your fix is to remove the "#, c-format"
 annotation.
 
 The second translation needs to have "%0" in it somewhere analogous to
 where it appears in the English, or the error messages will not make
 sense in context.
 
 I am attempting to get a more flexible mechanism for format-string
 handling put into gettext, but Bruno so far refuses to even consider
 it.  (Thank you for bringing this case to my attention, it illustrates
 a situation where both the existing choices are wrong.)
 
 zw

fortran/5558: Error in POT file

[martin]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1170 bytes --]


>Number:         5558
>Category:       fortran
>Synopsis:       Error in POT file
>Confidential:   no
>Severity:       critical
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Thu Jan 31 13:06:01 PST 2002
>Closed-Date:
>Last-Modified:
>Originator:     martin@v.loewis.de
>Release:        gcc-20020128.pot
>Organization:
>Environment:

>Description:
There are atleast two messages in the gcc catalog which are marked as c-format, but are not intended to be used with printf. As a result, msgfmt rejects the French translation. The specific examples are

#: f/lex.c:427
#, c-format
msgid "Hex escape at %0 out of range"
msgstr "échappement hexadécimal à %0 hors gamme"

#: f/lex.c:460
#, fuzzy, c-format
msgid "Escape sequence at %0 out of range for character"
msgstr "séquence d'échappement hors gamme pour le type caractère"
>How-To-Repeat:
Compile the current potfile with the given French translations (via msgfmt -c -v).
>Fix:
Please add a comment of /* xgettext:no-c-format */ just above the line that contains these messages.
>Release-Note:
>Audit-Trail:
>Unformatted: