Bug 191829
| Summary: | [stage] textproc/groonga: Update to version 4.0.4 | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Product: | Ports & Packages | Reporter: | tkato432 | ||||||||
| Component: | Individual Port(s) | Assignee: | John Marino <marino> | ||||||||
| Status: | Closed FIXED | ||||||||||
| Severity: | Affects Only Me | ||||||||||
| Priority: | --- | ||||||||||
| Version: | Latest | ||||||||||
| Hardware: | Any | ||||||||||
| OS: | Any | ||||||||||
| Attachments: |
|
||||||||||
provides staging. Miwi wants ports fury PRs automatically Created attachment 144978 [details]
textproc_groonga.diff
Return to pool I accidentally built this on i386 jail on amd64 (FreeBSD 10) and it failed. I have not tried redports yet. Have you tested this on i386? Another issue: It spews out a ton of warnings at least on FreeBSD10. I've been trying to figure out how to turn off. The warns seem to repeat over and over giving the appearance the compiler is stuck in a look but then it finally moves on. If you have any ideas (e.g. disable -Wall) i would be interested. I can reproduce this on redports: https://redports.org/buildarchive/20140731164931-48565/ 6x pass, 2x fail Fails on 10/i386 and 11/i386 The numerous warnings seem to a clang thing. There is some code in configure that checks specifically for gcc and clang but it wouldn't work for ports. I think it's also curious that the build takes far longer on F10 and F11 than F8 and F9 (2 to 3 times longer) Created attachment 145211 [details]
textproc_groonga.diff
I usually verify ports on 8.4/i386 and HEAD/amd64 only since common
troubles may happen on either of these combinations. An error only seen
on i386 with clang would be quite exceptional.
As far as I can see, disabling optimization is the only condition to
finish building this port flawlessly on 10.0/i386 with stock compiler.
Since expr.c consumes enormous amount of memory when compiling with
optimization, perhaps it reveals some shortcomings of clang on 32-bit
platforms if available free memory is insufficient. gcc appears to be
somewhat better at this point.
okay, so you're recommending gcc=any for i386 in any cases. Sounds reasonable to me. A commit references this bug: Author: marino Date: Fri Aug 1 13:38:31 UTC 2014 New revision: 363708 URL: http://svnweb.freebsd.org/changeset/ports/363708 Log: textproc/groonga: Upgrade version 3.0.0 => 4.0.4 and stage Note that gcc is required for i386 - clang/i386 will coredump on FreeBSD 10 and 11. PR: 191829 Submitted by: Ports Fury Changes: head/textproc/groonga/Makefile head/textproc/groonga/distinfo head/textproc/groonga/files/ head/textproc/groonga/pkg-install head/textproc/groonga/pkg-plist done! |
Created attachment 144600 [details] textproc_groonga.diff