On January 7, 2004 at 14:40, Jeremy Gebben wrote:
Here's a patch against nmh-1.1RC2 to make it compile on cygwin.
Cool. Sorry I never got around to sending you what I did.
Your stuff looks cleaner. You can see what I did at
<http://groups.google.com/groups?selm=1s0x8.108092%24CH1.93656%40sccrnsc02&oe=UTF-8&output=gplain>
In order of ascending nastiness, here's what I changed to make it
go:
-make a .PHONY dependency for the install target.
Note, this is a non-standard make feature. However, it may
be irrelevant to other platforms.
However, to make it standard, the common method of forcing a target
to be built evertime is to do something like:
install: _FORCE
... install commands here
_FORCE:
I.e. _FORCE is an empty target. You add it to the dependency list
of any target you always want built.
-cygwin doesn't provide a usable timezone variable in <time.h>,
it defines a function, char* timezone(void) instead. Supposedly,
defining timezonevar before including <time.h> will #define timezone
to the right type, but this causes <sys/time.h> to break
because it has a declaration for "struct timezone". Sigh. I've added
tests to configure.in that check if timezone is actually usable
as a long int, and if not it checks _timezone and uses that instead.
After I did my mess, I noticed through later searching about _timezone.
I tried to compute such things manually by comparing GMT time with
local time.
Note, dtime.c and dtimep.c had duplicate code (not sure if that
is still the case). The timezone stuff could be made into
a shared utility function.
+1 for adding the patch.
--ewh
_______________________________________________
Nmh-workers mailing list
Nmh-workers(_at_)nongnu(_dot_)org
http://mail.nongnu.org/mailman/listinfo/nmh-workers