Bug 239740
| Summary: | factor out hardcoding of "clangXYZ" in devel/qt5-qdoc | ||||||
|---|---|---|---|---|---|---|---|
| Product: | Ports & Packages | Reporter: | Mark Linimon <linimon> | ||||
| Component: | Individual Port(s) | Assignee: | freebsd-kde (group) <kde> | ||||
| Status: | Closed FIXED | ||||||
| Severity: | Affects Some People | CC: | adridg | ||||
| Priority: | --- | ||||||
| Version: | Latest | ||||||
| Hardware: | Any | ||||||
| OS: | Any | ||||||
| Bug Depends on: | |||||||
| Bug Blocks: | 239736 | ||||||
| Attachments: |
|
||||||
Perhaps the second time is the charm in terms of getting the portname correct. A commit references this bug: Author: adridg Date: Sun Aug 11 17:37:58 UTC 2019 New revision: 508658 URL: https://svnweb.freebsd.org/changeset/ports/508658 Log: Make devel/qt5-qdoc use LLVM_DEFAULT instead of hard-coded version Having the version hard-coded makes it hard to chase LLVM updates; having a different LLVM version from the rest of the stack -- in particular from what mesa uses -- means building an extra LLVM just for parsing documentation from Qt modules. linimon@ suggests using an easily-spotted variable in PR 239740 but we can go one better and just use the LLVM that's there. PR: 239740 Submitted by: linimon Changes: head/devel/qt5-qdoc/Makefile Let's just use LLVM_DEFAULT like sensible people (to avoid building an extra, different, compiler alongside the one that mesa needs anyway). Thanks for reporting! (In reply to Adriaan de Groot from comment #3) > Let's just use LLVM_DEFAULT I didn't ask for that because I haven't finished testing it yet :-) It's taking some time to run all the changes from 239736. > I didn't ask for that because I haven't finished testing it yet :-)
Now it's on my hat, not yours, and AFAICT qdoc is fine with it (also Qt upstream said it's ok, on IRC). So we're good (and 239736 is ever-so-slightly-smaller).
|
Created attachment 206392 [details] patch to devel/qt5-doc/Makefile There are a number of places in the ports tree where the string "clang<version" is called out. When doing the work to identify stale versions of clang tree-wide, these instances cause extra work. Refactor this usage into LLVM_VER. Note: this refactoring causes portlint to complain but IMHO this complication is worth tolerating to help ease maintainability.