Summary: | databases/redis-devel: fix build with GCC-based architectures | ||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Product: | Ports & Packages | Reporter: | Piotr Kubaj <pkubaj> | ||||||||||||
Component: | Individual Port(s) | Assignee: | Sergey A. Osokin <osa> | ||||||||||||
Status: | Closed FIXED | ||||||||||||||
Severity: | Affects Only Me | CC: | linimon, ndowens04, pkubaj | ||||||||||||
Priority: | --- | Flags: | bugzilla:
maintainer-feedback?
(osa) |
||||||||||||
Version: | Latest | ||||||||||||||
Hardware: | Any | ||||||||||||||
OS: | Any | ||||||||||||||
Attachments: |
|
Hold off on this one for now, compiler: breaks mips Created attachment 197588 [details]
Fixes build on mips, and possibly powerpc64
Test this patch on powerpc64, At first it was not building on mips, but this patch fixes mips issue atleast
(In reply to Nathan from comment #2) Yes, that works as well. Hi Piotr, thanks for the report. Would you mind to raise this issue and send the patch to upstream? (In reply to Sergey A. Osokin from comment #4) Please use my patch, as it won't break mips, but fix powerpc64 (In reply to Sergey A. Osokin from comment #4) I don't think upstream cares about GCC 4.2 :) (In reply to Nathan from comment #5) Since the patch is related to legacy gcc-specific architectures, would be great to update it to avoid intervention to other modern systems. (In reply to Sergey A. Osokin from comment #7) All I know to target is the ones that are reported to fail, hence why I only put powerpc64 with the USES=compiler:* If others fail without that compiler, could add them as well. I'm just trying to not break mips, while fixing one. If it needs to be added, as last example, it should be last resort and atleast a 2+:1 fix/broken ratio. If you know other ARCHs that need gcc, let me know and I will add them for the ARCHs. (In reply to Nathan from comment #8) powerpc, powerpcspe and sparc64 also need GCC. (In reply to Sergey A. Osokin from comment #7) Please, don't say "legacy". powerpc64 is very alive both as an architecture (with release of POWER9) and as a FreeBSD port (with recent addition of support for POWER8 and POWER9). It's far from legacy. Nathan, I see no reason to patch the source code with the patch on a platform where everything works fine. This is why I've recommended to update the patch to be more specific on affected platform(s). (In reply to Piotr Kubaj from comment #10) Piotr, as you can see I've said legacy regading outdated version of GCC, not a platform. Created attachment 197611 [details] Fix GCC based archs, keep mips working databases/redis-devel: Fix build on gcc based platforms PR: 231780 Submitted by: Nathan <ndowens@yahoo.com> This worked on mips and should work on others Created attachment 197612 [details] Fix GCC based archs, keep mips working databases/redis-devel: Fix build on gcc based platforms PR: 231780 Submitted by: Nathan <ndowens@yahoo.com> Add missing patch Thanks for the patch. In case this patch is for some specific platforms only, why we need to apply it for all other platforms? +.if ${COMPILER_TYPE} == clang +EXTRA_PATCHES= ${PATCHDIR}/extra-patches-src_lzf__d.c +.endif (In reply to Sergey A. Osokin from comment #15) We can apply it only for the platforms that are broken with this port, but it doesn't harm to apply it everywhere and will simplify Makefile. (In reply to Piotr Kubaj from comment #16) Would you mind to update the patch then. Thanks in advance. (In reply to Sergey A. Osokin from comment #17) The general patch that is applied everywhere has already been posted: https://bugs.freebsd.org/bugzilla/attachment.cgi?id=197588&action=diff (In reply to Piotr Kubaj from comment #18) Hi Piotr, It looks like I'm missing somesothing from your reply. Could you please explain a reason to use that patch on a platform where everything works just fine without the patch. Thanks. (In reply to Sergey A. Osokin from comment #19) I wrote in #16 and you asked me to update the patch appropriately: "We can apply it only for the platforms that are broken with this port, but it doesn't harm to apply it everywhere and will simplify Makefile." This patch does no harm on platforms that already work and fixes build on platforms that fail to build this port. (In reply to Piotr Kubaj from comment #20) I don't think I can agree with this. Again, I see no reason to apply a patch when it fixes nothing or just because it fixes something for other platforms. This is why I've asked to provide a more specific patch for the port and as far as I see the patch hasn't been updated yet. Created attachment 198305 [details]
v3
OK, that was a misunderstanding. Is this patch ok for you? It works for me on powerpc64 and amd64.
A commit references this bug: Author: osa Date: Thu Oct 18 22:20:52 UTC 2018 New revision: 482383 URL: https://svnweb.freebsd.org/changeset/ports/482383 Log: Fix build on GCC-based architectures. PR: 231780 Changes: head/databases/redis-devel/Makefile Committed, thanks! |
Created attachment 197567 [details] patch To build this port on architectures that still use GCC in base, you have to add USES=compiler:c11. Tested on amd64 and powerpc64.