Bug 210400
| Summary: | [maintainer][patch] www/node: remove erroneous post-configure step | ||||||
|---|---|---|---|---|---|---|---|
| Product: | Ports & Packages | Reporter: | Bradley T. Hughes <bhughes> | ||||
| Component: | Individual Port(s) | Assignee: | Adam Weinberger <adamw> | ||||
| Status: | Closed FIXED | ||||||
| Severity: | Affects Many People | CC: | adamw | ||||
| Priority: | --- | Keywords: | patch | ||||
| Version: | Latest | ||||||
| Hardware: | Any | ||||||
| OS: | Any | ||||||
| Attachments: |
|
||||||
I haven't had a chance to look at the source yet. Do you know whether that patch had any basis?
In general we don't want /usr/local hardcoded. Does it need to be subbed s|/usr/local|${LOCALBASE| ?
I think the basis was to try and remove the -I/usr/local/include that had been causing the problem. Personally, I feel that patching generated output isn't the right way to fix this... it feels too brittle and heavy handed. I'd rather find out how to make the upstream configure generate the right output. If we cannot do that, I'd like to patch the generator itself, and submit those patches upstream. FWIW, /usr/local is not hardcoded anywhere in the port. The /usr/local/* paths are getting pulled in via pkgconfig. Committed. Thanks, Bradley. A commit references this bug: Author: adamw Date: Mon Jun 20 20:09:53 UTC 2016 New revision: 417178 URL: https://svnweb.freebsd.org/changeset/ports/417178 Log: Revert part of r417063 that wasn't supposed to be committed. It removed a -I/usr/local/include, but it's present elsewhere in the CFLAGS so it doesn't change the final product. PR: 210400 Submitted by: maintainer (Bradley T. Hughes) Changes: head/www/node/Makefile |
Created attachment 171595 [details] patch from git, apply with patch -p1 An erroneous post-configure step snuck into the www/node/Makefile after the last PR before I could review the patch. This patch removes it.