Paul Eggert <[EMAIL PROTECTED]> wrote:
> I ran into a problem building CVS coreutils on Debian stable, since
> coreutils does not use some of the string.h-related modules, but
> Debian string.h #defines some of the symbols that string.h therefore
> attempts to #define to foo_is_unportable, and the #defines collide.
>
> I worked around my particular problem by installing the patch proposed
> below.  But I see further problems in this area.  For example,
> mountlist uses strstr in a system-dependent way where the ASCII-only
> system strstr is just fine.  A program that uses mountlist shouldn't
> need to use the gnulib strstr if it doesn't want to.  But with gnulib
> string_.h as it is, it will have to use gnulib strstr.

Thanks!  You beat me to it.

> How about if we remove the
> strstr_is_unportable__use_gnulib_module_strcasestr_for_portability
> symbols?  They do solve a problem, but I'm not sure it's worth
> the porting hassle of continuing to support them.
>
> Maybe there's a better way to address the same problem?

I hope so.  I like the safety net.
How about a c-strstr module, analogous to c-strtod?
i.e., strstr as if it were operating in the C locale.
It'd end up using vendor-supplied strstr in most cases.


Reply via email to