| 1 |
On Mon, 17 Dec 2007 07:17:32 +0000 |
| 2 |
Steve Long <slong@××××××××××××××××××.uk> wrote: |
| 3 |
> > PMS describes EAPIs 0 and 1, and will describe any future EAPIs once |
| 4 |
> > they're agreed upon. |
| 5 |
> > |
| 6 |
> It does a bit more than that. |
| 7 |
|
| 8 |
Mmmm, nope. Unless you mean "has an introduction page"... |
| 9 |
|
| 10 |
> And like I said, that's extraneous to Gentoo needs, and is in fact |
| 11 |
> only really needed for Paludis, not pkgcore and certainly not portage |
| 12 |
> development. |
| 13 |
|
| 14 |
It's entirely needed for Portage development. Without PMS, no change to |
| 15 |
Portage is allowed to break any existing ebuild, no matter how stupid |
| 16 |
what that ebuild is doing. |
| 17 |
|
| 18 |
> The last Council made that clear when they took the PMS in-house due |
| 19 |
> to the lack of progress: |
| 20 |
|
| 21 |
Which is kind of weird, given that there's plenty of progress... Also |
| 22 |
kind of weird that no-one in said Council or anywhere else has bothered |
| 23 |
to tell me about this in-house copy of the repository... |
| 24 |
|
| 25 |
> <wolf31o2|work> Gentoo has no need for a PMS if we're only supporting |
| 26 |
> portage... it was written pretty much exclusively to allow external |
| 27 |
> package managers to be on the same page as portage |
| 28 |
|
| 29 |
Untrue. |
| 30 |
|
| 31 |
> <kingtaco|work> it's possible that any PMS is of primary use for |
| 32 |
> external projects and perhaps we don't need to involve ourselves |
| 33 |
|
| 34 |
Possible. Not true. |
| 35 |
|
| 36 |
> <robbat2> i see the goal of PMS as allowing external PMs to be |
| 37 |
> supported in Gentoo |
| 38 |
|
| 39 |
Partially true, in that it's one of the goals. |
| 40 |
|
| 41 |
> I wonder how far you'd get with trying to, say, supplant rpm in |
| 42 |
> RedHat, or apt in debian. Surely they must be crying out for a |
| 43 |
> next-gen PM? |
| 44 |
|
| 45 |
You mean "I wonder how far you'd get with trying to, say, supplant |
| 46 |
dselect in debian". |
| 47 |
|
| 48 |
Oh, right. |
| 49 |
|
| 50 |
You'll note that RPM and APT have proper specifications, and aren't |
| 51 |
defined in terms of any particular implementation. |
| 52 |
|
| 53 |
> Oh yeah, you need the Gentoo devs to maintain ebuilds or |
| 54 |
> paludis won't work. |
| 55 |
|
| 56 |
Which has what to do with anything? (It's also untrue, but hey, that's |
| 57 |
a separate issue.) |
| 58 |
|
| 59 |
> <kingtaco|work> no, the discussion on -dev ml defines eapi bumps |
| 60 |
|
| 61 |
You're taking things out of context. Discussion on -dev (or rather, |
| 62 |
bugzilla) is used to decide what goes in an EAPI. The EAPI itself is |
| 63 |
described formally elsewhere. |
| 64 |
|
| 65 |
> Having a spec isn't an issue: the issue is having it developed as a |
| 66 |
> mainstream Gentoo project, with open discussion. |
| 67 |
|
| 68 |
Anyone who knows what they're talking about is more than welcome to |
| 69 |
discuss things. Anyone who doesn't, which clearly includes you at this |
| 70 |
point, is more than welcome to go and do the necessary research. |
| 71 |
|
| 72 |
-- |
| 73 |
Ciaran McCreesh |
Archives