Marius Mauch, développeur gentoo qui participe en particulier au développement de portage, sera présent cette année au FOSDEM pour une conférence au sujet de ce gestionnaire de paquetages.
http://www.fosdem.org/2005/index/speakers/speakers_mauch(…)
Vous pouvez lui poser ici vos questions jusqu'au Dimanche 30 Janvier 2005 au matin. L'équipe du FOSDEM les intégrera alors dans l'interview qui lui sera envoyée puis publiée sur le site du FOSDEM.
# Pôv' Marius, il a pas de question...
Posté par tgl . Évalué à 3.
- I think that since i know Gentoo (~2.5 years) I hear about "portage-ng". Rumors went from the idea of a C++ rewrite to the one of a multi-languages rewrite with a Prolog dep solver. Approx 1 year ago, there has been a road-map embryo published and a call for features requests, but I don't remind it going up to a spec draft. And afaik, the in-progress work is more a big refactoring of the existing Python code to make it much more modular and to separate all the non-frontend code from emerge. So, what is (will be), for you, this mysterious "portage-ng" (if anything), and do all devs agree on their vision of that beast?
- There are currently several way to write helper tools (like the ones from gentoolkit) for Gentoo. Some use perl or bash and access directly to portage files (the ebuild tree and/or the /var/db/pkg repending on the needs), other use the python modules from portage (portage.py and friends), other use the classes from gentoolkit (gentoolkit.py), etc. What would be your advice to people who want to write such tools? Is there currently an API that is stable enough, or should they wait, or...?
- Here is a list of some frequently requested features for Portage that I'm sure you've also seen several times either on b.g.o or on f.g.o. Could you tell if and when you think they will be (or should be) implemented? You can answer by "RTFM" / "next minor release" / "next major release" / "never, but if contributed" / "definitly never!", or discuss more if you want. Sure, feel free to add or drop items at your convenience...
* Per-packages CFLAGS
* Per-category USE flags, ACCEPT_KEYWORDS and CFLAGS
* "/etc/portage/package.env"
* User-defined packages sets (world subsets)
* Multiple portage configurations support to ease building binaries for several targets on a same host
* Same as above, but with a robust cross-compiling support
* "einfo" and friends logging and reporting
* Multiple slots installation of a same package-version
* Dynamic libraries tracking for safe package ugrade or uninstallation
* Real backwards dependencies
* Per-slot depend atoms
* With-USE-flags depend atoms
* Fast (cached) package search in emerge
* On-demand ebuild downloading, or partial tree rsyncing
* "emerge --security"
* "emerge -u --severity={any, minor-features, major-features, critical-fixes, security-fixes}"
* GPG signed portage tree
* Per-ebuild USE flag doc hints
* USE flags groups
* Interactive emerge, to change USE flags, etc.
* Multiple ebuilds tree syncing
* Automatic kernel modules rebuilding
* Progress bar for compilations
* Recursive package categories, or multiple logical categories per package
* Random speed improvements (SQL engine to replace /var/db/pkg, C/C++ rewritings of some time-critical code, Psyco support, etc.)
And now some less technical questions:
- What are the top 3 reasons for you being a Gentoo user? And a Gentoo developer?
- What are the 3 major improvements you would like to see on Gentoo in 2005? And are they realistic?
- What is the most beautiful piece of code you've ever seen in the amount of bash/python/etc. that Gentoo devs have produced so far? And the worst one?
- How would you qualify the Gentoo user base? And are you happy with that as a dev?
[^] # Re: Pôv' Marius, il a pas de question...
Posté par pvdabeel . Évalué à 2.
Portage-ng is a project created by Daniel Robbins, Nicholas Jones (Current portage lead) and myself. its goal was to implement functionality that required non-trivial changes to the Portage design. I'll also attend Fosdem 2005
Best regards,
Pieter Van den Abeele
[^] # Re: Pôv' Marius, il a pas de question...
Posté par tgl . Évalué à 2.
# Troll
Posté par tgl . Évalué à 2.
# source VS binary VS source+binary
Posté par Sebastien . Évalué à 2.
What would be for you the improvements (if any) one could implement to make gentoo a binasource or source-ary distribution ? So (hopefully) the best of both worlds would meet in the perfect distro.
# Source Mage vs Gentoo
Posté par Jiel (site web personnel) . Évalué à 2.
Suivre le flux des commentaires
Note : les commentaires appartiennent à celles et ceux qui les ont postés. Nous n’en sommes pas responsables.