Created attachment 251652 [details] Patch for mmg * Add patches to remove git build dependency * Remove "make test" target, broken due to missing test files, this also removes bash build dependency * Minor changes to Makefile to more closely follow Porters Handbook Poudriere testport OK 14.1-RELEASE (amd64)
Created attachment 251653 [details] Patch for mmg Correct patch
Please forget me! If you're bored, instead of criticizing or deprecating ports that are still working, make yourself useful: there are hundreds of PRs to process, and hundreds of ports without a maintainer to adopt.
Forget? Pulling in git because of a few lines of code which returns nothing useful is wasteful and using git to determine version is error-prone in general. Unit tests are not usable at all, "4% tests passed, 300 tests failed out of 312" most fail instantly because test data files are not packaged. That's not really working even by a stretch.
(In reply to Thierry Thomas from comment #2) Thierry, sorry to chime in here - what you are writing here is not how we have seen ports and quality in the past, and also should not. Telling volunteers and co-committers who do care about ports quality where they should focus instead, or insinuating they were bored when they in fact do care, is not helpful at all and not contributing to a collegial and progressive atmosphere. I also do not see reasons to feel personally offended, Daniel seems to having wanted to improve ports quality, and the reasonable recourse is either considering and integrating the contribution, or giving up port maintainership.
He is right about git, but it was not my priority (building a full-featured Gmsh), and you know, a port is rarely perfect the first time. Many tests are failing because the required data are not included in the main tarball, and it was in my plan to make them available separately later. Being harassed by counter-productive people is not fun: I just released it.
Thierry, I fail to see how anyone could consider a request for improvement as "being harrassed". I do not see any personal texts here. What I would however very much like to see is that ports that offer "make test" pass the test on the mainstream OS and ARCH - so we know when stuff breaks at "build-time" and not at run-time, which I see as improving the overall user experience. And I am not sure that I like too much if someone adds a new port that isn't ready without it being marked experimental or anything. What's your urge? June 30 approaching for 2024Q3? Then quality is also urgent... I appreciate that different people in a large project may have different standards or let's rather say priorities as to when to consider something "done", but then also you seem to be frustrated with other ports - what does this have to do with a newly added mmg port? Re-enabling tests later once you had had added those tests is a matter of usually one Makefile line, so that cannot be the actual reason. (Thierry may not see a copy from BZ directly any more, I will make sure he sees my comment.)
A commit in branch main references this bug: URL: https://cgit.FreeBSD.org/ports/commit/?id=c222dbace702197660e280c3fe0fa76ecb3c69f0 commit c222dbace702197660e280c3fe0fa76ecb3c69f0 Author: Daniel Engberg <diizzy@FreeBSD.org> AuthorDate: 2024-06-30 07:40:15 +0000 Commit: Daniel Engberg <diizzy@FreeBSD.org> CommitDate: 2024-06-30 07:40:20 +0000 science/mmg: Improve port * Add patches to remove git build dependency * Remove "make test" target, test files aren't bundled and also also remove bash dependency * Patch perl script(s) using USES= shebangfix * Adjust Makefile to more closely follow Porters Handbook PR: 279935 science/mmg/Makefile | 12 +++++------- science/mmg/files/patch-CMakeLists.txt (new) | 14 ++++++++++++++ science/mmg/files/patch-git__log__mmg.h (new) | 9 +++++++++ 3 files changed, 28 insertions(+), 7 deletions(-)