Eric Blake wrote:
> shouldn't all the rpl_*printf variant
> declarations be given a printf attribute when available so that users of
> the *printf-posix modules still get the benefits of gcc format string
> checking?
Good idea. I'm applying this:
2007-03-10 Bruno Haible <[EMAIL PROTECTED]>
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Matthew Woehlke on 3/8/2007 9:28 AM:
>
> (We already know this doesn't work: if configure doesn't define
> HAVE_LONG_LONG_INT, then the resulting stdint.h breaks system headers.)
>
>
> Without also telling gnulib not to use 'long long',
The patch from 2007-01-14 to avoid "./configure; make" during --create-testdir
did most often not work. This should fix it.
2007-03-10 Bruno Haible <[EMAIL PROTECTED]>
* gnulib-tool (func_create_testdir): Treat MOSTLYCLEANFILES like
CLEANFILES. Put spaces in each line of $cleane
Simon Josefsson wrote:
> Consider making it possible to not use this workaround, but
> rather fail with a compilation error. I wouldn't want code that
> invoke abort in a library, for example.
Makes sense. The Solaris 10 include files contain comments like this:
* Even though the contents of th
Eric Blake wrote:
> 2007-03-10 Eric Blake <[EMAIL PROTECTED]>
> and Matthew Woehlke(tiny change)
>
> * lib/stdint_.h (int64_t, uint64_t): Don't undefine if 64-bit
> types are not being provided by gnulib.
> (GL_INT64_T, GL_UINT64_T): New witnesses of whether gnulib 64
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Bruno Haible on 3/10/2007 8:25 AM:
> Eric Blake wrote:
>> 2007-03-10 Eric Blake <[EMAIL PROTECTED]>
>> and Matthew Woehlke(tiny change)
>>
>> * lib/stdint_.h (int64_t, uint64_t): Don't undefine if 64-bit
>> types are n
