Paul Eggert <[EMAIL PROTECTED]> wrote: > Eric Blake <[EMAIL PROTECTED]> writes: > >> Thanks for the report. Which platform, which compiler, and which version >> of coreutils was this? If it is still present in coreutils 6.2, we would >> like to get it fixed before 6.3. > > Yes, likewise. > > Looking at the implementation, it seems like there are some obvious > checks to add, e.g., that one can seek to a 4 GiB boundary in a > mkstemp file. Also, there are some naming problems, since some > stdlib.h have "#define mkstemp ..." in some situations, and this runs > afoul of config.h's "#define mkstmep rpl_mkstemp". > > I don't know whether the following patch will fix his problem (we need > more info for that) but I don't see how it could hurt, so I installed > it into gnulib. > > 2006-09-20 Paul Eggert <[EMAIL PROTECTED]> > > * modules/mkstemp (Depends-on): Add extensions, so that > mkstemp is visible on some platforms. > (Makefile.am): Add mkstemp.h to EXTRA_DIST. > (Include): Change to "mkstemp.h" from <stdlib.h>. > * lib/mkstemp.h: New file, since some standard headers > #define mkstemp. > * lib/mkstemp.c: Revamp to put the !_LIBC code together. > Include "mkstemp.h". > Make the _LIBC code resemble glibc original more, > e.g., use K&R style. > * lib/stdlib--.h: Include mkstemp.h. > * m4/mkstemp.m4 (gl_FUNC_MKSTEMP): Require AC_SYS_LARGEFILE. > Check that large offsets work. Modernize Autoconf usages. > Prefer "yes" to mean a good thing rather than a bad. > Don't put "#define mkstemp" in config.h, as this might interfere > with standard system headers that "#define mkstemp mkstemp64".
That looks fine. Thanks!
