Re: autoconf, sendmail, and pthreads

[Steven G. Johnson]

On Tue, 19 Dec 2000, Lars Hecking wrote:
>  Can anyone comment on ACX_PTHREAD?

rssh wrote:
> It does not work good with solaris and freebsd.

When you have a bug report, contacting the author is generally a good
idea...

As a matter of fact, I updated ACX_PTHREAD after feedback a month or three
ago that it had problems on Solaris (the problem was that Solaris has
broken pthread stubs in libc that confused the tests), and it should work
there now.  I haven't heard of any problems on FreeBSD, and in fact I have
specifically tested it on FreeBSD systems in the past.

rssh wrote:
> I. e. if anybody will touch this macro, than  merging into acx_pthread
> stuff from RSSH_CHECK_PTHREADS from corbaconf 
> [ http://www.corbaconf.kiev.ua ] would be a good idea.

Thanks for the link, I've downloaded this code.  It looks like it only
handles three cases: a "-mt" flag, a "-pthreads" flag (but only on
FreeBSD, as it uses system-name lookups), and a "-lposix4" library (it
misses many common cases that are handled by acx_pthread, e.g. -lpthread
or -lpthreads).  However, I should add -mt and -lposix4 tests to my macro,
so this is useful to know.

-mt is for Sun Workshop C, according to the comments in the code.  In my
experience, Solaris has used -pthread or -lpthread; is this for an
older/newer system?  Is -mt equivalent to -pthread, or does it do anything
else?  (Unfortunately, I don't have an account on a Solaris machine at the
moment.)

What system uses -lposix4 for its threads library?  And why do you test
for it by looking for the "nanosleep" function instead of something in the
pthreads API?

(It is easy enough to add tests for -mt and -lposix4 into acx_pthread
without knowing what they are for, but explanatory comments are nice, and 
testers are better.)

Cordially,
Steven G. Johnson