Re: National Language Support Library

["Julius Enarusai" <jenarusa@us.ibm.com>]

>I would suggest that these routines by used conditionally as
>we should like try to use XPG4 routines provided by the C library
>(whenever available).  When "builtin" XPG4 routines are used, they
>should be used through macros which keep our symbols from clashing
>with "real" XPG4 routines.  That is, some header (ldap_nl_types.h)
>should define catgets() to ldap_pvt_catgets().  Our code should
>use catgets() as if it has the real thing.
>

We are currently looking into this and would replace any calls to catgets() with appropriate macros where possible.

>I would note that the routines would likely have to be placed
>in -llber...

We were thinking of delivering a separate library which would contain the routines provided. However, -llber or any other library is perfectly OK.

>I would suggest one start localization effort with -lldap's
>ldap_err2string() and other strings generated by the API and
>returned normally to applications.  This is what "users"
>see first.

I completely agree with you. If we provide this translation to ldap_err2string(), it would make it transparent to users (i.e. no code change on the part of users).

>The localization should, of course, be driven by the environment.
>Applications can manage this environment on behalf of the
>use...  our code should localize to whatever locale is returned
>by setlocale( LC_MESSAGES, NULL ).

>I note that OpenLDAP may still have some pieces of code dependent
>on running in the "C" locale.

Yes. The library uses setlocale(LC_MESSAGES, NULL) to determine the user's local language.



Julius Enarusai
 Directory Services,
 IBM Linux Technology Center
 Phone: 856-722-6569
 Tie Line: 242-6569
 Fax:      856-722-5937
 Cell:      410-340-8020