Advice about /usr/ports/math/gmp

Torbjorn Granlund tege at
Tue Jan 7 11:32:11 UTC 2014

I duly regret to have tried to save some time for the gmp ports
maintainer.  I am slapping my trollish hands!  (I even mispellt a word,
gasp!)  By all means, try upgrading, run into clang bugs and m4 bugs.
(The limits.h bugs won't happen for ports builds, only for plain
compiles by end users.)

The m4 bug has been around for a long time, no *release* is unaffected.
I have been told 9-STABLE has it fixed.  I know the imminent 10 has it
fixed.  Newer ports is supposed to work on older systems, right?

Last time I tried (recently) clang had some bug(s) (neither of which
fully isolated nor reported).  How can I be sure it is a compiler bug?
Why didn't I report it?  I am sure it is a bug in clang, did enough to
check that.  I didn't report it because I used a lot of time for my
latest fbsd bug report (amd64/183397), and 10 is shipping with it.

I have been with FreeBSD since 1.1.  It has been my main development
platform for GNU work ever since.  In the last period, I have been
unlucky enough to see a mounting number of show-stopper problems with
FreeBSD.  I don't recognise my beloved FreeBSD, nor do I recognise its

I will have nothing more to say about this.  Scream and shout about this
reply if you feel inspired, but please cut me out from that.

If I--privately or via the appropriate gmp list--sometimes in the future
get a reasonably polite inquiry about integrating GMP 5.2.x into
FreeBSD's ports, I will repeat my advise (!) but I or some other GMP
hacker will certainly make an effort trying to help if the ports hacker
insists in doing the integration.

Bye, bye.

More information about the freebsd-ports mailing list