Paul D. Smith writes:
> Bruno, do you have any thoughts on how we could _safely_ auto-detect a
> "good" gettext implementation? I think it would be satisfactory to
> detect a version of GNU gettext, and brand all other versions "bad",
That's good enough, yes. You can detect GNU gettext by check
%% =?ISO-8859-1?Q?Fran=E7ois_Pinard?= <[EMAIL PROTECTED]> writes:
fp> Bruno Haible <[EMAIL PROTECTED]> writes:
>> The default in make-3.79 -- borrowed from libit-0.7 -- is to *always*
>> use the included gettext!
fp> I'm ready to change the default, given that some clues are
fp> auto-
%% Bruno Haible <[EMAIL PROTECTED]> writes:
bh> And the definition of _GNU_SOURCE in "make.h" comes too late, because at
bh> the point where it is defined, a system header file has already been
bh> included ( from "gettext.h"). Fix below. In particular,
bh> "make.h" must always be the fir
Bruno Haible <[EMAIL PROTECTED]> writes:
> The default in make-3.79 -- borrowed from libit-0.7 -- is to *always*
> use the included gettext!
I'm ready to change the default, given that some clues are auto-configured
to discover that the installed gettext is better than the included one.
Includin
Hi,
Compiling make-3.79 with a current glibc snapshot. glibc-2.2 will have
new features in gettext: automatic conversion of the translation to the
right charset, and plural handling.
1) The default in all normal programs using AM_GNU_GETTEXT is to use the
included gettext only if the system doe
Hi,
Compiling make-3.79 with a current glibc snapshot.
Autoconfiguration found out that the function 'strsignal' is available.
But the header files declare it only when compiling with -D_GNU_SOURCE.
(Which is ok, because 'strsignal' is by no means a standardized function.)
So we get a warning:
