Re: [PATCH 1/2] open_memstream-tests: new module

2010-04-26 Thread Eric Blake
On 04/24/2010 04:21 AM, Bruno Haible wrote: > Eric Blake wrote: >> +open_memstream-tests: new module >> +* modules/open_memstream-tests: New file. > > I would mark it as an unportable test, like this: > > Status: > unportable-test Sure, although that designation can be removed with m

Re: [PATCH 1/2] open_memstream-tests: new module

2010-04-24 Thread Bruno Haible
Eric Blake wrote: > + open_memstream-tests: new module > + * modules/open_memstream-tests: New file. I would mark it as an unportable test, like this: Status: unportable-test The reason is that there are (so far) 7 types of stdio implementations: - FreeBSD, NetBSD, OpenBSD, DragonF

Re: [PATCH 1/2] open_memstream-tests: new module

2010-04-23 Thread Eric Blake
On 04/23/2010 05:21 PM, Paul Eggert wrote: > Eric Blake writes: > >> + ASSERT (STREQ (buf, "hello my world")); >> + ASSERT (len = strlen (buf)); > > These two tests, plus some others, assume that open_memstream allocates > a buffer in which the output is NUL-terminated. This isn't a portable

Re: [PATCH 1/2] open_memstream-tests: new module

2010-04-23 Thread Paul Eggert
Eric Blake writes: > + ASSERT (STREQ (buf, "hello my world")); > + ASSERT (len = strlen (buf)); These two tests, plus some others, assume that open_memstream allocates a buffer in which the output is NUL-terminated. This isn't a portable assumption (though it is often true, which I expect is

[PATCH 1/2] open_memstream-tests: new module

2010-04-23 Thread Eric Blake
Test passes on Linux and cygwin 1.7, but fails everywhere that open_memstream is not implemented. * modules/open_memstream-tests: New file. * tests/test-open_memstream.c: Likewise. Signed-off-by: Eric Blake --- ChangeLog|6 modules/open_memstream-tests | 12 ++