[Groff] Re: Question concerning Czech man pages in UTF-8

[Mike Fabian]

Werner LEMBERG <address@hidden> writes:

>> Running the following commandline 
>> 
>>     zcat /usr/share/man/cs/man1/man.1.gz | groff -mandoc -Tutf8 - | less
>> 
>> in an UTF-8 locale in an UTF-8 capable terminal results in the display
>> of wrong characters in a Czech man page:
>> 
>>     http://www.suse.de/~mfabian/screenshots/czech-man-page-utf-8.png
>> 
>> But the following commandline 
>> 
>>    zcat /usr/share/man/cs/man1/man.1.gz | (LC_ALL=cs_CZ.ISO-8859-2 groff 
>> -mandoc -Tlatin1 - ) | iconv -f iso-8859-2 -t utf-8 | less
>> 
>> appears to work fine:
>> 
>>     
>> http://www.suse.de/~mfabian/screenshots/czech-man-page-utf-8-with-help-of-iconv.png
>> 
>> Why is it necessary to force groff to run in cs_CZ locale and use
>> -Tlatin1 (or -Tascii8) and convert the result to UTF-8 instead of
>> using -Tutf8 directly?
>
> Currently, groff only supports UTF8 for output, not for input.  This
> will change as soon as groff's conversion to Unicode has been
> implemented.

OK, thank you. As a temporary workaround until groff supports UTF-8
for input, I changed /usr/bin/nroff a little bit.  In case of
ja_JP.UTF-8 locale or cs_CZ.UTF-8 locale I call groff in the
respective non-UTF-8 locale and convert the result to UTF-8 with
iconv.

The /usr/bin/nroff I'm currently using is attached.

nroff
Description: Binary data

-- 
Mike Fabian   <address@hidden>   http://www.suse.de/~mfabian
睡眠不足はいい仕事の敵だ。