1 |
-----BEGIN PGP SIGNED MESSAGE----- |
2 |
Hash: SHA1 |
3 |
|
4 |
On 08/01/2012 08:29 AM, Fabian Groffen wrote: |
5 |
> In a little less than two weeks from now, the council will meet |
6 |
> again. This is the time to raise and prepare items that the council |
7 |
> should put on the agenda to vote on. |
8 |
> |
9 |
> Please respond to this email with agenda items. Please do not |
10 |
> hestitate to repeat your agenda item here with a pointer if you |
11 |
> previously suggested one (since the last meeting). |
12 |
> |
13 |
> The agenda for the next meeting will be sent out a week before the |
14 |
> meeting by the meeting's chairman. |
15 |
> |
16 |
> Please respond to gentoo-project list, if possible. |
17 |
> |
18 |
> |
19 |
|
20 |
As per devmanual |
21 |
http://devmanual.gentoo.org/ebuild-writing/common-mistakes/index.html |
22 |
build log should always be verbose. |
23 |
|
24 |
A tracker has been added: |
25 |
https://bugs.gentoo.org/show_bug.cgi?id=429308 |
26 |
|
27 |
However, it does not make much sense to fix all of this per-package, |
28 |
since we can change the behavior of certain build systems. For cmake |
29 |
this has been mostly fixed via cmake-utils.eclass. |
30 |
|
31 |
For autotools based build systems we need to modify econf as it has |
32 |
already been done in: |
33 |
http://git.overlays.gentoo.org/gitweb/?p=proj/portage.git;a=commit;h=1cc39de72ac5311db748341ef9183586556719d9 |
34 |
(add --disable-silent-rules if the configure script has this option) |
35 |
and then reverted again, cause PMS guys intervened this will cause |
36 |
breakage, because './configure --help' will be unconditionally invoked |
37 |
by econf (however this is already the case for EAPI=4, but not for 0-3). |
38 |
|
39 |
reason for this commit was: |
40 |
In order to have consistent behavior that follows this gentoo policy |
41 |
accross the whole tree we should always try to fix this above |
42 |
per-package level, either on eclass or PM level. |
43 |
Otherwise we will have to file many bugs for this. This has actually |
44 |
been requested by another dev who was annoyed by my bugreports. |
45 |
|
46 |
========== |
47 |
The council should vote on the following: |
48 |
1. apply that portage commit again and add that information to PMS |
49 |
retroactively. |
50 |
2. If point 1 is declined, then vote on applying this to EAPI=4 |
51 |
retrocatively since this one already calls './configure --help'. |
52 |
========== |
53 |
|
54 |
Why point 1 should be declined as per PMS guys opinion: |
55 |
It may break ebuilds and changes older EAPIs behavior. |
56 |
|
57 |
Why point 2 should be declined as per some random guys opinion: |
58 |
It changes older EAPIs behavior. |
59 |
|
60 |
Why point 1 should be accepted IMO: |
61 |
If this only applies to EAPI=4 then all affected packages using EAPI=3 |
62 |
or lower would either need an EAPI-bump or append |
63 |
- --disable-silent-rules to econf. |
64 |
Both is per-package level and will result in many many bugs being |
65 |
filed again (and I predict many of them will just be ignored, because |
66 |
devs will be annoyed by them). |
67 |
If an ebuild that uses EAPI=3 breaks, because econf simply calls |
68 |
'./configure --help', then it is already borked and should be fixed, |
69 |
cause it's using a handcrafted configure script with econf which was |
70 |
designed for autotools based build systems. |
71 |
This is not a strong reason to avoid this retroactive change which |
72 |
will bring us consistent behavior regarding autotools based build systems. |
73 |
|
74 |
Why at least point 2 should be accepted IMO: |
75 |
There is no technical argument against it. No one could think of any |
76 |
kind of breakage this might cause to EAPI=4 + we have at least a |
77 |
retroactive fix for this one to get SOME consistent behavior for the |
78 |
tree. It will take years until the majority of ebuilds will have |
79 |
migrated to EAPI=5. |
80 |
Don't rely on future EAPIs to fix current problems within the tree. |
81 |
|
82 |
|
83 |
You can follow part of the discussion on: |
84 |
https://bugs.gentoo.org/show_bug.cgi?id=379497 |
85 |
|
86 |
|
87 |
Don't discuss this here please. I just tried to sum it up, so I might |
88 |
have missed something. |
89 |
-----BEGIN PGP SIGNATURE----- |
90 |
Version: GnuPG v2.0.19 (GNU/Linux) |
91 |
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ |
92 |
|
93 |
iQEcBAEBAgAGBQJQI8TBAAoJEFpvPKfnPDWzM5IH/2bzr4Z8HwgCUw7mihvp2VDg |
94 |
Ld26sgtn0s/MpaqAeA+6LjgNG26bAmk19bNYhd7Atm1LoYTsPFzLuiYYQDv6wGZN |
95 |
Bf+3/n5YYXgPAwkD9q9q/h1fTHl45CSycd73gL8a3/3jVcEt9HX+9Qp7g7ZnD/NQ |
96 |
dWD1c3snb5I9iUbOZZw75IauJZujxN4AnWxp2Z1ncGl5xHrjIgOY0RPbHFDWyUyj |
97 |
tOPBh1SQIEPmkjagcb9vHXrSQMiTj4+TJH8xqN3yPR7ZYu76TijV8WOnBjUsrHpW |
98 |
QxOEzaShd1mHpc2tWn1vi/stvdZQttxXGihc+d8HNFJYJ3z9kULfRVGyOuS3qkg= |
99 |
=NVAx |
100 |
-----END PGP SIGNATURE----- |