Re: gawk-3.1.3-mbstate_t.patch for hpux

[Michael Elizabeth Chastain]

Hi Stepan,

mec> gawk really requires mbstate_t.  Look at strncasecmpmbs for example.  So
mec> it's reasonable to add a HAVE_MBSTATE_T test.

stepan> I think it's better to use the AC_FUNC_MBRTOWC() macro instead.
stepan> It checks whether both mbrtowc() *and* mbstate_t are available.

I saw that macro and I chose not to use it.

I like to keep things orthogonal and atomic.  It's possible for a
platform to have mbrtowc() function and no mbstate_t type, and it's
possible for some applications to work properly on such a platform
(although gawk is not such an application).  So I think these should be
separate tests.

I am new to autoconf, so maybe I just have bad taste.

mec>    * configure: Regenerate.
mec>    * configh.in: Regenerate.
stepan> I beleive there is no nee to put the following to your patch:

That's cool.  The maintainer can strip them out and rerun
autoconf themselves, of course.  I need to have a patch that I can
apply to gawk 3.1.3 as released and then build gawk, so I included
them in my patch.

stepan> Michael, are you willing to rework the patch according to my
stepan> suggestions and to verify the it really works on hpux?

Well, no.  I've got a lot of things to do and this is volunteer work,
like much of Gnu work.  It's not even my primary volunteer work.  So I
can't afford to spend much time on it.

If someone else wants to send me modified patches to test I would be
happy to run them through an hpux machine.  Or take the interesting
parts of my patch and put them into the gawk cvs repository and point me
at a cvs repository or a tarball and I'll test it.

(By the way, I use HP's test drive program.  Anyone else can get access
to the same machines, although it takes some time to ftp in all the
source you need to actually do work).

Michael C