<<< Date Index >>>     <<< Thread Index >>>

Re: national chars - libiconv not used?



Hi Alain,
        
I've spent a nice few days by looking for a way how to find problem in a locale definition (before I opened this thread), but unsuccessfully. The locale was installed by our information system vendor. But I believe it was installed by copying appropriate files from old K260, ux 10.20. I have not noticed other locales problem; but, on the other side, mutt is the first thing working harder with locales which I'm trying to compile.

And for the command:
hpuh:/usr/lib/nls/loc # localedef -f charmaps/IBM852.cm -i src/cs_CZ.IBM-852.src /tmp/cs_CZ.IBM852

Processed LC_CTYPE

Processed LC_COLLATE

Processed LC_MONETARY

Processed LC_NUMERIC

Processed LC_TIME

Processed LC_MESSAGE
sh: /opt/ansic/bin/cc:  not found.
localedef [ERROR]: FILE: src/cs_CZ.IBM-852.src, LINE: 730, CHAR: 1
Unable to exec /usr/bin/sh to process localedef intermediate files.

hpuh:/usr/lib/nls/loc #
I have not HP cc ansi compiler. I tried numerous way how to trick this, but with no luck.


Thanks,
        Michal


Alain Bench wrote:
 On Friday, July 7, 2006 at 8:16:32 +0200, Michal Hajek wrote:


isprint(167) = 128
iswprint(167) = 128
mbrtowc returns 1, and converted char 0xA7 to wc U+00A7


    OK: I Googled HP manpages and such: It's not failing, but working as
designed. The HP-UX 11 wide character type is explicitly stated as not
being Unicode, but some locale dependant value. And as those 80-9F
characters are all recognised printable, why the hell does Mutt require
+LOCALES_HACK and the additional patch, and why does Ncurses show "~A"
as if they were control chars? I'm dry. For the moment.


    I also Googled the roman8 CODESET problem: Nothing. I mean nobody
reports such problem, to the contrary many examples show good behaviour.
Now, manpages say that HP locales are architecture specific: PA-RISC,
Itanium, 6502, 32 bits, 64 bits... And locale/localedef commands have
options to list/create only an arch locale, and files have arch
directories. Perhaps could you dig that? Also found a list of
preinstalled locales: cs_CZ.IBM-852 is not listed. How was it created?

| $ localedef -f IBM852 -i cs_CZ  /tmp/cs_CZ.IBM-852
| value of field `int_curr_symbol' in category `LC_MONETARY' does not 
correspond to a valid name in ISO 4217
| no output file produced because warning were issued

    Woops! ;-)

| $ LC_ALL=cs_CZ locale int_curr_symbol
| CZK


Bye!    Alain.


--
--------------------------------------------------------------
Michal Hajek                   mailto:hajek@xxxxxxxx
Sprava NIS                       http://www.nspuh.cz
NsP Uherske Hradiste          phone : voice +420 572 529 204
J. E. Purkyne 365                     fax   +420 572 551 014
686 68 Uherske Hradiste  Czech Republic
--------------------------------------------------------------