Hi, On Wed, Jul 08, 2026 at 10:25:37AM +0200, Marc Espie wrote: > > In light of that, I wonder if it would be better to remove the > > `UPDATE_PLIST_ARGS += -t ...` stuff from texmf/Makefile, and have > > update_plist_hints.py be the source of truth for what to filter out. > > That way we'd have the logic in one place. > > Not a good idea. -t will look in the ports tree, and thus be automatically > up-to-date wrt the actual files in the port, as opposed to having to handcode > them in your script. I've been thinking about this some more. `UPDATE_PLIST_ARGS += -t <pkgpath>` filters out exact paths already included in the PLIST of <pkgpath>. update_plist_hints.py filters out whole tex packages. Thats what I was doing here: ``` conflict_pkgs = ["asymptote", "latexmk", ...] conflict_pkg_files, conflict_symlinks = \ collect_files(allspecs(conflict_pkgs, include_deps=False), db) comment_files = CONFLICT_FILES | conflict_pkg_files | commented_docs_files ``` By filtering at the tex-package-level, we remove: - The files for the whole tex package, as it appears in the TLPDB. - But, *regardless of where the files are installed*. Looking at our asymptote port, it installs a load of tex files into `share/texmf-local/tex/latex/asymptote`, but texlive (if we let it) would install those same files under `share/texmf-dist/tex/latex/asymptote`. So, if I understand correctly, that's why `UPDATE_PLIST_ARGS += -t graphics/asymptote` isn't sufficient to filter out the duplicated asymptote files: they are installed in different paths. Therefore, it seems to be more robust for me to filter out files from tex packages that are already separately ported at the update_plist_hints.py level, since `UPDATE_PLIST_ARGS -t` will only remove them if the files are installed at the same path. What do others think? -- Best Regards Edd Barrett https://www.theunixzoo.co.uk
No comments:
Post a Comment