Hello, I would like to raise a discussion on the topic of the /etc/portage/env files support hack, implemented in the profile.bashrc file in gx86 base profile [1].
I am working on the portage-side implementation of the feature and I would like to avoid double-sourcing of the override files. Although that could be implemented through checking for the existence of new Portage private variable, which would be used to pass the override directory path to the bash part of Portage, I would rather see the hack removed ASAP. Why? Because it should never had been added there in the first place. It is not only a hack -- it is a hack silently changing the behaviour of all package managers out there (assuming they source profile.bashrc) for all users. The hack bypasses the whole concept of PM package stabilization - it 'injects' the bash code into every version of Portage, pkgcore and Paludis (I guess). A code targeted at Portage, I'd add. And it is poorly documented too. I wasted a lot of time trying to find the code related to the feature in the Portage code and searching for some references before I realized that thing has nothing to do with Portage source code! To sum up, I think that the hack should be removed ASAP -- even before Portage starts supporting the feature the 'correct' way. It could be temporarily replaced with a check for existence of '/etc/portage/env' directory, printing a short warning that the hack has been disabled and user should either wait for Portage release supporting the feature (or upgrade his Portage version, if the removal is going to happen after the related rc-release) or implement it /etc/portage/bashrc-side. [1] http://sources.gentoo.org/viewcvs.py/gentoo-x86/profiles/base/profile.bashrc?view=markup -- Best regards, Michał Górny <http://mgorny.alt.pl> <xmpp:mgo...@jabber.ru>
signature.asc
Description: PGP signature
