Message ID | 20230225185544.5152-1-ludo@gnu.org |
---|---|
Headers | show |
Series | Package cleanups: Guile, GnuTLS, ncurses | expand |
Hey! I passed a long list of ‘--add-header="X-Debbugs-Cc: …"’ flags to ‘git format-patch’ in an attempt to reach the whole ‘core’ team + Andreas, but it turns out that only the last ‘--add-header’ option had an effect (or rather it overrode all the previous occurrences). We really need to provide reliable instructions for teams. Anyway, here’s a patch set! https://issues.guix.gnu.org/61790 Ludo’. Ludovic Courtès <ludo@gnu.org> skribis: > Hello! > > Here’s a patch series that’s hopefully not too late, cleaning up core > packages and thus introducing a world rebuild. > > I realize this will be annoying to people like Andreas who have been > testing things much higher in the stack (Chromium!). What we could do > is have ci.guix build more than the “core” subset (the current setting); > it could build ‘etc/release-manifest.scm’, which contains things up > to IceCat. > > Thoughts? > > Ludo’. > > Ludovic Courtès (5): > gnu: guile: Remove input labels. > gnu: gnutls: Remove Guile dependency. > gnu: gnutls: Deprecate 'gnutls-latest'. > gnu: gettext: Remove input label. > gnu: ncurses: Rewrite using gexps. > > gnu/packages/base.scm | 4 +- > gnu/packages/gettext.scm | 5 +- > gnu/packages/guile.scm | 60 ++++---- > gnu/packages/ncurses.scm | 301 +++++++++++++++++++-------------------- > gnu/packages/tls.scm | 62 +++----- > 5 files changed, 200 insertions(+), 232 deletions(-) > > > base-commit: cb3332808f42beb4c08970ed792875ee145b715c
Am Sat, Feb 25, 2023 at 11:14:16PM +0100 schrieb Ludovic Courtès: > Here’s a patch series that’s hopefully not too late, cleaning up core > packages and thus introducing a world rebuild. I am not too worried about a world rebuild, but more about introducing new build failures, leading to indefinite delays. At first glance, the patches do not look like they have a high risk, but who knows?
Maybe it would be safer to leave them out for the moment, and to start a feature branch "core" right after the core-updates merge? In any case, having a CI job would be very welcome, so that people could more easily test packages closer to the leaves than the root of the package graph. Andreas
Hello! Andreas Enge <andreas@enge.fr> skribis: > I am not too worried about a world rebuild, but more about introducing new > build failures, leading to indefinite delays. At first glance, the patches > do not look like they have a high risk, but who knows? […] > Maybe it would be safer to leave them out for the moment, and to start > a feature branch "core" right after the core-updates merge? There’s always a risk of course, though I think it’s pretty low in this case (I built everything up to guile{,2.2}-gnutls, which includes like 8h spent in ‘nss’ tests). Removing the dependency of GnuTLS on Guile is the most desirable part; the rest is more cosmetic. We can delay that until after ‘core-updates’ too, but that might take a while since <https://ci.guix.gnu.org/eval/228947/dashboard> stills shows quite a bit of red. :-) Thoughts? Tough choices! Ludo’.
Am Sun, Feb 26, 2023 at 06:16:46PM +0100 schrieb Ludovic Courtès: > There’s always a risk of course, though I think it’s pretty low in this > case (I built everything up to guile{,2.2}-gnutls, which includes like > 8h spent in ‘nss’ tests). Removing the dependency of GnuTLS on Guile is > the most desirable part; the rest is more cosmetic. Okay, if you tried it locally, then I think we can take the risk. Andreas
Andreas Enge <andreas@enge.fr> skribis: > Am Sun, Feb 26, 2023 at 06:16:46PM +0100 schrieb Ludovic Courtès: >> There’s always a risk of course, though I think it’s pretty low in this >> case (I built everything up to guile{,2.2}-gnutls, which includes like >> 8h spent in ‘nss’ tests). Removing the dependency of GnuTLS on Guile is >> the most desirable part; the rest is more cosmetic. > > Okay, if you tried it locally, then I think we can take the risk. Alright, pushed as 4895846197bd445dd68c832364c667dea4337f8c. Thank you for your understanding! :-) Ludo’.