Created attachment 253734 [details] Patch for tiff Compile and runtime tested on FreeBSD 14.1-RELEASE (amd64) (make, make check-plist, make test) Poudriere testport OK 13.3-RELEASE (amd64) Poudriere testport OK 14.0-RELEASE (i386) Poudriere testport OK 14.0-RELEASE (amd64) Poudriere testport OK 14.1-RELEASE (amd64) A "mini-exp" run on my end looks fine
It is a bit strange that you use INSTALL_MAN to install docs in post-install-DOCS-on. Is this a copy-paste typo?
It's what Porters Handbook suggests https://docs.freebsd.org/en/books/porters-handbook/book/#install-macros
This is peculiar, I never noticed that. It also says "it does not compress anything" which is false for man pages, but OK. Another point is switching the build to CMake. I personally don't mind, but other people may not welcome this change. I failed to find out what buildsystem the upstream recommends.
Hi, I'd like to request an exp-run Best regards, Daniel
(In reply to Gleb Popov from comment #3) It builds faster and provides .cmake files which upstream projects makes use of so I don't see this as a regression and distros such as Alpine Linux have already migrated to CMake.
Might also add that it speeds up unit tests (which also runs more tests than Autotools) and the majority of consumers also uses CMake. Autotools: make 14.55s real 34.53s user 6.56s sys make test: 6.32s real 3.89s user 2.63s sys CMake: make: 9.42s real 32.59s user 2.48s sys make test: 1.81s real 6.03s user 2.41s sys
Upstream needs to explicitly recommend using cmake; the README only mentions autotools. The "builds faster" and all other arguments are irrelevant. Not every distro chooses to use cmake (Arch Linux still uses autotools)
You missed the first sentence? This file is just a placeholder; the entire documentation is now located as reStructuredText in the doc directory. https://libtiff.gitlab.io/libtiff/build.html or https://gitlab.com/libtiff/libtiff/-/blob/v4.7.0/doc/build.rst
It's still a README at the end of the day and it shall be treated as such. That part of the documentation seems ambivalent about using either. CI also appears to run both. In which case, switching from one to the other is only really worth when accompanied by an API/ABI break, which these release notes mention none.
Please switch back to configure/make. Symbol versionning is known to be totally broken with tiff built with cmake.
(In reply to Antoine Brodin from comment #10) Just a note - "maintainer-feedback -" means "no feedback" rather than "patch not accepted".
(In reply to Antoine Brodin from comment #10) Can you elaborate on this as doesn't seem to be case in 4.7.0 ?
Might as well add that the symbol lists https://reviews.freebsd.org/P648
New failure logs: https://pkg-status.freebsd.org/gohan04/data/140amd64-default-foo/2024-09-26_19h10m22s/logs/dcmtk-3.6.8.log https://pkg-status.freebsd.org/gohan04/data/140amd64-default-foo/2024-09-26_19h10m22s/logs/mandelbulber-2.32.log https://pkg-status.freebsd.org/gohan04/data/140amd64-default-foo/2024-09-26_19h10m22s/logs/vtk8-8.2.0_5.log https://pkg-status.freebsd.org/gohan04/data/140amd64-default-foo/2024-09-26_19h10m22s/logs/smoldyn-2.73_1.log
A commit in branch main references this bug: URL: https://cgit.FreeBSD.org/ports/commit/?id=0bdf5887f0d2cc9f14ea0cc10ed9e371028dd620 commit 0bdf5887f0d2cc9f14ea0cc10ed9e371028dd620 Author: Daniel Engberg <diizzy@FreeBSD.org> AuthorDate: 2024-10-09 19:12:09 +0000 Commit: Daniel Engberg <diizzy@FreeBSD.org> CommitDate: 2024-10-09 19:43:25 +0000 graphics/tiff: Update to 4.7.0 and switch to CMake Changelog: https://gitlab.com/libtiff/libtiff/-/releases/v4.7.0 PR: 281639 Approved by: desktop (fluffy) via Matrix Exp-run by: antoine graphics/tiff/Makefile | 32 ++++++--- graphics/tiff/distinfo | 6 +- .../files/patch-cmake_LinkerChecks.cmake (new) | 11 +++ .../tiff/files/patch-libtiff_CMakeLists.txt (new) | 11 +++ .../patch-tools_unsupported_Makefile.in (gone) | 10 --- graphics/tiff/pkg-plist | 81 ++++++++++++++++++++-- 6 files changed, 124 insertions(+), 27 deletions(-)