-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 On 23/11/12 09:28 AM, Rich Freeman wrote: > On Fri, Nov 23, 2012 at 9:15 AM, Ian Stakenvicius <[email protected]> > wrote: >> .. For certain things, I think it would be very beneficial for >> this to be true (other dev's welcome to touch) across the tree. >> Maybe if there is enough general support for it, we should change >> our default of "never touch a maintainer's package without >> permission of the maintainer/herd", to "OK to touch unless >> package metadata explicitly requests not to" ...? And we can put >> a tag in the metadata to indicate this (or even to indicate what >> other dev's can and can't touch -- ie, can touch *DEPEND, can >> bump EAPI, cannot add features, cannot bump)? > > Honestly, I like the maintainer/herd system - it promotes some kind > of consistency and accountability. If everybody just goes poking > at random ebuilds anytime they want to then that will tend to lead > to chaos. >
I'm not suggesting to abandon that, just augment it a little. There are dev's that want strict do-not-touch-my-stuff control, and dev's that don't really care. Defining as such in metadata would keep a persistent record. I can think of two specific examples where this would be an advantage: #1 - the init-script-license issue. When I filed all of those bugs, there were a few dev's that said to me "Do what you want to fix the LICENSE= on your own", many others didn't but i'm guessing that didn't mean they actually explicitly desired to control LICENSE=. Similarly, I have absolutely no problem at all of someone fixes LICENSE= in any of my packages -- I set them properly as best I could and I try and watch out for changes, but if there's someone that knows better I say "just do it." #2 - sub-slots and slot-operators. Adoption of this will go a lot faster if the maintainers of libraries had free reign to update *DEPEND in rdeps when necessary. Some of this already happens, but having permission to adjust *DEPEND be explicitly listed would make it go quicker still and not cause the inevitable arguments that it always seems to.. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.19 (GNU/Linux) iF4EAREIAAYFAlCvivYACgkQ2ugaI38ACPCSxwEAhudOC/pqJhcDPj1LErF8C2f1 bPAYrfcdNRCnovPSS2sA/jhkGjgkPcBFIM/m4uMKq8hVmHqw5RDb86pljpJz+37P =ihwa -----END PGP SIGNATURE-----
