Re: Re: news item: Introduction of app-alternatives/coreutils
Sam James <[email protected]>
| Newsgroups | gmane.linux.gentoo.devel |
|---|---|
| Organization | Gentoo |
| Message-ID | <[email protected]> |
Ionen Wolkens <[email protected]> writes: > On Sat, Mar 21, 2026 at 11:36:18PM +0500, Anna Vyalkova wrote: >> On 2026-03-21, moltonel 3x Combo wrote: >> > Hi, >> > >> > This news item can also be reviewed at >> > https://github.com/gentoo/gentoo-news/pull/5 >> > The corresponding code PR is at https://codeberg.org/gentoo/gentoo/pulls/89 >> > This is my first time posting here and I'm not a Gentoo dev, apologies >> > if I made mistakes. >> > >> > >> > Title: Introduction of app-alternatives/coreutils >> > Author: Vincent de Phily <[email protected]> >> > Posted: 2026-03-04 >> > Revision: 1 >> > News-Item-Format: 2.0 >> > >> > A new app-alternatives/coreutils ebuild is available, giving the >> > choice between sys-apps/coreutils and sys-apps/uutils-coreutils as the >> > main coreutils (mkdir, ls, etc) implementation. The default will >> > remain the gnu implementation for the foreseeable future. >> > >> > The update should be transparent to most users, read on for further >> > details. >> >> As it has been pointed out, there's barely any need for a news item. It >> looks like an advertisement that most users won't find useful. > > Admit I'm not fond of the visibility bit either, I'd rather that this > goes under the radar (like virtual/zlib did) so a potentially large > amount users don't try switching on a whim after reading news (even if > we mask USE=uutils). > > I don't really care about the rust or licensing bits, it's more about > potential compatibility issues that may still need to be ironed out. > Even moreso when (for Gentoo) build systems are involved. > > ... albeit a transition without a news warning does also worry me, > app-alternatives switches haven't always been the smoothest given > users also do all sort of things that may make it go badly, and then > this is quite a core package. So not quite sure what I think of this. > This is what I'm torn on and I don't yet know. I'd prefer to not have to do one indeed given I think it will encourage people or give some legitimacy to it being a "reasonable option". >> To me, "rewriting coreutils in Rust" looks like a move against copyleft. >> If authors of uutils did not intend it, companies like Canonical switch >> to uutils not because it's better but because they hate GPL. But >> whatever, Gentoo is about choice. >> >> > ## Choices >> > >> > If uutils-coreutils is available for your arch and profile, you >> > can select it by setting `app-alternatives/coreutils -gnu uutils` in >> > your packages.use. >> > >> > If you want to postpone or revert the transition, mask the >> > corresponding ebuild versions: >> > >> > ``` >> >>=sys-apps/coreutils-9.10-r1 >> >>=sys-apps/uutils-coreutils-0.6.0-r1 >> > app-alternatives/coreutils >> > ``` >> > >> > ## Preparation and troubleshooting >> > >> > The new version of sys-apps/coreutils and sys-app/uutils-coreutils >> > install their files with a prefix (`g` for gnu, and `uu-` for uutils). >> > Afterwards, app-alternatives installs symlinks with standard names. >> > >> > Portage preserves some critical binaries, which will trigger a >> > harmless file collision warning (assuming the default `protect-owned` >> > portage feature is enabled) when app-alternatives/coreutils is >> > installed for the first time. >> > >> > To reduce the window during which some utils will not be available at >> > the standard location, make sure that app-alternatives/coreutils is >> > merged straight after the new sys-apps/coreutils, by running >> > `USE="gnu -uutils" emerge -v1 app-alternatives/coreutils` before >> > updating the rest of the system. >> > >> > In the unlikely event of an incomplete update, keep in mind you can >> > still use the prefixed names. If you somehow need to manually fix some >> > symlinks, use the command below and consider reporting the issue: >> > >> > ```bash >> >> Markdown syntax is not supported for gentoo news items. Please keep it >> formatted in plain text. >> >> > # Critical utils that portage should have preserved >> > export U1="[ arch basename cat chgrp chmod chown cp date dirname echo env expr >> > false head hostname id install ln ls mkdir mkfifo mktemp mv printf >> > pwd readlink rm rmdir sleep sort stat tail test touch tr true uname >> > uniq whoami" >> > # Utils that might be missing until app-alternative installs >> > export U2="b2sum base32 base64 basenc chcon chroot cksum comm csplit cut dd >> > df dir dircolors du expand factor fmt fold hostid join kill link >> > logname mknod nice nl nohup nproc numfmt od paste pathchk pinky pr >> > printenv ptx realpath runcon seq shred sha1sum sha224sum sha256sum >> > sha384sum sha512sum shuf split stdbuf stty sum sync tac tee timeout >> > truncate tsort tty unexpand unlink users vdir wc who yes" >> > for t in $U2; do >> > gtest -e /usr/bin/g$t -a ! -e /usr/bin/$t && gln -sf g$t /usr/bin/$t >> > gtest -e /bin/g$t -a ! -e /bin/$t && gln -sf g$t /bin/$t >> > done >> > ``` >> > >> > The current implementation of app-alternatives/coreutils does not >> > switch shell completions, so `cp` uses gnu completions, `uu-cp` uses >> > uutils completions, and `gcp` does not complete. This will be >> > addressed progressively, in collaboration with upstreams. >> > >> > ## Further reading >> > >> > [0] https://wiki.gentoo.org/wiki/Project:Base/Alternatives >> > [1] >> > https://www.gentoo.org/support/news-items/2022-12-27-alternatives-introduction.html >> > [2] https://bugs.gentoo.org/961068 >>
signature.asc
(application/pgp-signature, 418 B)
-----BEGIN PGP SIGNATURE----- iQEBBAEWCgCpFiEEJaa7iN2bdkxrVUHCc4QJ9SDfkZAFAmm/OBobFIAAAAAABAAO bWFudTIsMi41KzEuMTIsMiwyXxSAAAAAAC4AKGlzc3Vlci1mcHJAbm90YXRpb25z Lm9wZW5wZ3AuZmlmdGhob3JzZW1hbi5uZXQyNUE2QkI4OEREOUI3NjRDNkI1NTQx QzI3Mzg0MDlGNTIwREY5MTkwDxxzYW1AZ2VudG9vLm9yZwAKCRBzhAn1IN+RkLNY AP90Rzxf9a7cAHZFiuK88bIoS7/55ou+X+Qff9Yyyoy+WgEA9ohCN/hVAH2wsJNe 8jCkK1BKzF7t4z8xLaKoG+iETAM= =FYdb -----END PGP SIGNATURE-----