[Bug 261255] math/nfft: Clean up port (WIP) - Help wanted
- In reply to: bugzilla-noreply_a_freebsd.org: "[Bug 261255] math/nfft: Clean up port (WIP) - Help wanted"
- Go to: [ bottom of page ] [ top of archives ] [ this month ]
Date: Wed, 19 Jan 2022 08:25:28 UTC
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=261255 --- Comment #1 from Fernando ApesteguĂa <fernape@FreeBSD.org> --- (In reply to Daniel Engberg from comment #0) Hi Daniel, I built the attached version of the port and I don't see any instances of fomit-frame-pointer in the logs: /usr/local/poudriere/data/logs/bulk/latest-per-pkg/nfft/3.5.2_2$ grep fomit-frame-pointer * /usr/local/poudriere/data/logs/bulk/latest-per-pkg/nfft/3.5.2_2$ I understand that those are the flags the port uses unconditionally regardless of C/LD FLAGS? On the issue of cexpl(). It is unfortunate that we don't have it in older versions of FreeBSD. nfft does not provide a "test" target to check for the correctness of the built artifacts, so I would mark the port BROKEN for versions not having cexpl(). math/openlibm seems a mature project, but still, it might lead to very subtle bugs to use different math libraries depending on the FreeBSD version. Since this is a leaf port, this shouldn't cause a cascade of problems. For curiosity I just did a git reset --hard a94f8e0af96a4182e9428d3d02f20c3d8325ecf7 and rebuilt nfft-3.5.1. It builts OK even in 12.2{amd64,i386}. Inspecting the code, cexpl() was already there, but I see that it is checked for its declaration in configure, so maybe somehow its usage was conditional but I can't see how. -- You are receiving this mail because: You are the assignee for the bug.