1 |
On Fri, 2008-01-11 at 09:59 -0500, Doug Klima wrote: |
2 |
> Mike Frysinger wrote: |
3 |
> > after dealing with m68k, mips, and the *-fbsd ports, i think we could do with |
4 |
> > a new state for profiles.desc. the new field would simply be "exp" to |
5 |
> > indicate that the profile is experimental and that qa tools should generally |
6 |
> > not issue warnings about them. so in repoman's default mode, you wouldnt get |
7 |
> > any warnings, but if you were to run it in full mode, you'd see stuff like |
8 |
> > normal. this is useful for new projects which are still in the process of |
9 |
> > merging (like *-fbsd, m68k, and any new hardware i get my hands on) and are |
10 |
> > not really ready for "dev" marking. there are plenty of warnings in packages |
11 |
> > right now due to these profiles being labeled as "dev" that are the sole |
12 |
> > problem of the keyword maintainer in question and not the package maintainer. |
13 |
> > -mike |
14 |
> > |
15 |
> I've very much been a proponent of properly listing out all of our |
16 |
> profiles in profiles.desc. This sounds very reasonable and logical. And |
17 |
> hopefully if the tools in question are coded properly, it should be |
18 |
> compatible with older versions until users upgrade. |
19 |
|
20 |
To go along with this, I've always wanted a few other minor changes. |
21 |
|
22 |
For one, a way to mark a profile as deprecated in profiles.desc so |
23 |
repoman doesn't scan it (currently, we remove tend to remove them from |
24 |
the list). The second would be a change to repoman that's more |
25 |
"invasive" in that it changes current behavior a good bit, but having |
26 |
repoman only scan "stable" profiles, by default, with options to scan |
27 |
the other types. I've always wanted to have *every* valid profile |
28 |
listed in profiles.desc so we can do things like have portage not allow |
29 |
someone to use a profile that isn't listed in profiles.desc (of course, |
30 |
overlay users crazy enough could do their own profiles.desc and it would |
31 |
be stacked with the in-tree one). The main problem with doing this has |
32 |
been the effect on repoman, since it scans every listed profile every |
33 |
time. I know that most of the profile selection tools out there already |
34 |
only show profiles that are listed in profiles.desc, so it wouldn't |
35 |
really be a change for them, but I think it would be useful elsewhere, |
36 |
too. All in all, having profiles.desc actually showing the status of |
37 |
all of the profiles would be great. |
38 |
|
39 |
> Good stuff Mike. |
40 |
|
41 |
Indeed. |
42 |
|
43 |
-- |
44 |
Chris Gianelloni |
45 |
Release Engineering Strategic Lead |
46 |
Games Developer |