#4176 closed enhancement (wontfix)
Minor enanchements at GMP Portfile
| Reported by: | mellon85@… | Owned by: | macports-tickets@… |
|---|---|---|---|
| Priority: | Normal | Milestone: | |
| Component: | ports | Version: | 1.0 |
| Keywords: | Cc: | gwright@… | |
| Port: |
Description
Now there is a powerpc variant and an optimizing CFLAGS environment variable. I also added a parallel make, it can really make compilation faster on dual processor machines and is harmless on single ones
Attachments (1)
Change History (4)
Changed 19 years ago by mellon85@…
| Attachment: | gmp-Portfile.patch added |
|---|
comment:1 Changed 19 years ago by mellon85@…
Is it right to have compile optimizing flags in the Portfile or it is left at the flag set by the user??
comment:2 Changed 19 years ago by gwright@…
| Resolution: | → wontfix |
|---|---|
| Status: | new → closed |
Hi Dario,
I've looked at the proposed patch. The configuration procedure already sets the -O2 optimization flag, as well as -mpowerpc. It is almost never necessary to set these flags explicitly in a package that uses GNU autotools properly. (The main reason for doing so is that the configure script is broken on a platform and needs help figuring out what to setting to use.)
The -funroll-loops might help---it is not set by default---but that should be tested by a timing comparison. It may be that cache locality is hurt by that option so it slows the library down, rather than speeding it up. If you can send me some benchmarks comparing the two cases, we can decide what to do.
Generally speaking, darwinports doesn't use parallel build (make -j2) because they can occasionally fail in hard-to-debug ways. The default make procedure should always be safe.
I have modified the port to run "make check" after building. Any changes made in the future should pass all of the tests.
Best Wishes, Greg
Portfile patch