other/5927: the PO files may contain full sentences rather than phrases

[=?utf-8?q?Nilg=C3=BCn=20Belma=20Bug=C3=BCner?= <nilgun at fide dot org>]

>Number:         5927
>Category:       other
>Synopsis:       the PO files may contain full sentences rather than phrases
>Confidential:   no
>Severity:       non-critical
>Priority:       low
>Responsible:    unassigned
>State:          open
>Class:          change-request
>Submitter-Id:   net
>Arrival-Date:   Tue Mar 12 13:26:01 PST 2002
>Closed-Date:
>Last-Modified:
>Originator:     Nilgün Belma Bugüner, gcc.tr.po translator
>Release:        3.1 (20020307)
>Organization:
home user
>Environment:
System: Linux nilgun.transsoft.home 2.4.12-ac3 #2 Sat Feb 9 18:51:27 GMT 2002 i686 unknown
Architecture: i686 
host: i686-pc-linux-gnu
build: gcc 3.1 20020307 (Red Hat Linux Rawhide 3.1-0.22)
configured with: ../configure --prefix=/usr --mandir=/usr/share/man --infodir=/usr/share/info --enable-shared --enable-threads=posix --disable-checking --host=i386-redhat-linux --with-system-zlib
/usr/lib/gcc-lib/i386-redhat-linux/3.1/cc1 -lang-c -v -D__GNUC__=3 -D__GNUC_MINOR__=1 -D__GNUC_PATCHLEVEL__=0 -D__ELF__ -Dunix -Dlinux -D__ELF__ -D__unix__ -D__linux__ -D__unix -D__linux -Asystem=posix -D__NO_INLINE__ -D__STDC_HOSTED__=1 -Acpu=i386 -Amachine=i386 -Di386 -D__i386 -D__i386__ -D__tune_i386__ help-dummy -quiet -dumpbase help-dummy -version --help -o /tmp/ccQgxvYm.s

>Description:
I am the Turkish translator for gcc. When translating gcc and testing 
the translation, I noticed that certain phrases, as-is, cannot be 
translated into Turkish correctly. I try to give examples below, but 
basically the same English phrase would translate into slightly 
different Turkish phrases depending on the context. It is impossible to 
figure out the context from the phrases itself and the result is 
incorrect Turkish in some cases.

I would like to request that complete sentences be used instead of 
incomplete phrases.  This is a problem we face whenever two separate 
strings are added programatically to make up what the end user sees.  

For example:

The following two strings are concatenated together in the program to 
make up a complete message visible to the end user.
"%s from incompatible pointer type"
"passing arg %d of `%s'"

The resulting string as printed during program execution is:
"warning: passing arg %d of `%s' from incompatible pointer type"
Translating these two phrases separately and concatenating them via the 
program does not result in an intelligible Turkish sentence.

Second example:
Two strings:

"passing arg %d of `%s'"
"%s discards qualifiers from pointer target type"

The resulting string as printed during program execution is:
"passing arg %d of '%s` discards qualifiers from pointer target type

This particular sentence ends up with a correct Turkish translation when 
the two strings above are translated separately into Turkish and 
concatenated via the program.

As you can see, it is impossible to guarantee a correct translation when 
phrases are put together by a program to make up a sentence for the end 
user. Translating these phrases separately and putting them together to 
make up a sentence may or may not result in a correct sentence in the 
translated language.

While putting in complete sentences results potentially in higher number 
of strings to be translated (as opposed to the above example where 
"passing arg %d of %s" phrase is common to both sentences and is 
translated just once), it actually is to be preferred because:
1. We guarantee an intelligible translated end result
2. It actually makes the job of translation easier for the translator. 
More translation is done, but trying to guess at the context and hunting 
for the context in the program source code is lessened.

>How-To-Repeat:
>Fix:
Can you please make changes in the gcc program source as needed so that 
the PO files contain full sentences rather than phrases which are put 
together by the program to make up complete sentences?
>Release-Note:
>Audit-Trail:
>Unformatted:
 X-send-pr-version: 3.113