1 |
Roy Bamford wrote: |
2 |
|
3 |
> On 2009.06.07 10:34, Ulrich Mueller wrote: |
4 |
>> >>>>> On Sun, 07 Jun 2009, Steven J Long wrote: |
5 |
>> |
6 |
>> > I'd just like to know what the implications would be for users if |
7 |
>> we |
8 |
>> > kept the .ebuild extension, and a new PMS were rolled out stating |
9 |
>> > that the mangler were allowed to find the EAPI without sourcing |
10 |
>> (and |
11 |
>> > giving the restrictions) once portage 2.2 was stable, or the |
12 |
>> ability |
13 |
>> > to handle this backported to 2.1.6, and issued in a release? |
14 |
>> |
15 |
>> Even if we do only a one-time change of the file extension, can we |
16 |
>> ever get rid of the old extension? Or are we then stuck with two |
17 |
>> extensions in the tree until the end of time? |
18 |
>> |
19 |
>> Let's assume for the moment that we change from ".ebuild" to ".eb". |
20 |
>> Then we obviously cannot change all ebuilds in the tree to ".eb", |
21 |
>> otherwise old Portage versions would see an empty tree and there |
22 |
>> would |
23 |
>> be no upgrade path. |
24 |
>> |
25 |
>> Or am I missing something? |
26 |
>> |
27 |
Sounds about right |
28 |
|
29 |
> |
30 |
> First, lets be clear that the upgrade path related problems are driven |
31 |
> by the EAPI change, not the .ebuild to .eb change. That serves only to |
32 |
> allow the new EAPI to be introduced immedately and hide the change from |
33 |
> older package managers |
34 |
> |
35 |
<snip> |
36 |
> To keep an upgrade path, package managers and their dependencies need |
37 |
> to use EAPI 0, 1 or 2. (EAPI 3 is not yet out) How far back should the |
38 |
> upgrade path extend? |
39 |
> |
40 |
Well given that EAPI 3 is not out, and that bash-4 is not even stable |
41 |
(and if anyone thinks we can rely on bash-4 in the next 6 months, they |
42 |
didn't learn anything from bash-3 imo) this sounds like it's a fair |
43 |
distance away. From discussion with Harring, EAPIs were not meant to |
44 |
come out very often; iirc he said something like maybe once a year. |
45 |
|
46 |
I concur with Duncan on a year, as you know. I appreciate you feel it |
47 |
should be longer, and am happy to go with whatever the consensus is. |
48 |
|
49 |
> As you suggest, even with .ebuild to .eb change.its not a clean break |
50 |
> with the past but would happen over time, with ebuilds for new package |
51 |
> versions being migrated to the new format. For example we would |
52 |
> have |
53 |
> foo-2.1-r1.ebuild |
54 |
> foo-3.0.eb |
55 |
> portage-2.3.ebuild |
56 |
> portage-3.0.eb |
57 |
> |
58 |
yuck. |
59 |
|
60 |
> Portage-2.3 will understand the later EAPI but will itself, use only |
61 |
> EAPI, 0, 1 or 2. |
62 |
> |
63 |
Just to be clear: portage-2.2 and later will be fine with any EAPI, with |
64 |
no change to any ebuilds, nor any new extension being needed. The |
65 |
change can easily be backported to 2.1.6, tho I suspect 2.2 will be |
66 |
stable fairly soon. |
67 |
|
68 |
> Thats just portage. The same reasoning applies to any other package |
69 |
> manager and there are at least three. This begs the question of how |
70 |
> friendly do we want to be to derivitive distros that use our tree but |
71 |
> their own package manager? |
72 |
> |
73 |
As friendly as we can be without hobbling our own development. Most of |
74 |
them appear to be fairly collaborative with Gentoo in any case. |
75 |
|
76 |
> Upgrade path issues need to be addressed in the GLEP. I will add |
77 |
> something like the above in the next version. |
78 |
> |
79 |
Wrt transitioning, can the eapi (PMS 5.2.2) and deprecated (5.2.3) files |
80 |
not be of use? I seem to recall the change from 2007.1 to 2008.0 for |
81 |
example; anyone using a deprecated profile can expect a massive warning |
82 |
the next time they try to do anything after sync'ing. |
83 |
|
84 |
Thanks again for your work; I appreciate that your time is valuable. |
85 |
|
86 |
Regards, |
87 |
Steve. |
88 |
-- |
89 |
#friendly-coders -- We're friendly but we're not /that/ friendly ;-) |