[Nmh-workers] nmh-1.3-RC1: restoring build functionality lost since nmh-1.0.4

[Russell Ruby]

The big burst of activity on the nmh mailing list, along with
the release of nmh-1.3-RC1 inspired me to upgrade from nmh-1.0.4 on
my desktop machine - a sparc5-170 running SunOS4.1.4.  (I bought that
sucker new and I'm gonna get my money's worth!   Besides, I'm
used to it.)

There were a few build and run glitches needing fixing,  but nmh-1.3-RC1
has now been chewing through mail just fine for the last few days.
I have submitted a couple of patches in the project "bugs" section so
far, but forgot to reference nmh-1.3-RC1 in the title of the second
submission.  That one restores the ability to build outside the
"top src" directory, along with restoring the automatic substitutions
for missing library functions.

I have more build and runtime patches for SunOS - but currently
they are all SunOS4.1.4 specific hacks as in:

#if defined(sun) && !defined(__svr4__) && !defined(__SVR4)
   ...
#else
  ...
#endif

Seems like making these general enough for submission would require
adding new specific tests to configure.in along with macro definitions.

Here are the problem areas for SunOS4.1.4 which had to be "fixed":

1) Although HAVE_TZSET and HAVE_STRUCT_TM_TM_GMTOFF are both set,
   HAVE_TZSET breaks the build,  undef'ing HAVE_TZSET and leaving
   HAVE_STRUCT_TM_TM_GMTOFF defined does build, but I haven't tested
   the result at run time yet.  undef'ing both builds and has been
   used for several days now.
   (sbr/dtime.c)

2) SunOS4.1.4 realloc does not like being fed a null pointer (sbr/utils.c)

3) SunOS4.1.4 MB_CUR_MAX returns zero when called in uip/scansbr.c, ouch.

4) an include of sys/ioctl.h in uip/termsbr.c   I don't remember why
   I added that.

5) SunOS sprintf returns a pointer to the produced string, not the
   count of characters.  (uip/whatnowsbr.c)

6) missing functions:  strerror, snprintf, and memmove
   

I think that was it.  Not all that much.


Thanks,

russell ruby