1 |
Jakub Moc wrote: |
2 |
> |
3 |
>>>Currently there are quite a few ebuilds in the tree that execute dodoc or |
4 |
>>>dohtml for files that do not exist. I think it would be nice to have |
5 |
>>>ebuilds die if this is the case. To not break current ebuilds this would |
6 |
>>>only happen with FEATURES="stricter". |
7 |
> |
8 |
> |
9 |
> Sigh... There are already bugs flowing in for TEXTRELs/executable stacks |
10 |
> checks implemented in recent portage versions. Some of these bugs are |
11 |
> completely INVALID or CANTFIX - emulation stuff, binary-only ebuilds, etc. |
12 |
> etc. What's the point of this breakage? Why are these QA checks fatal, |
13 |
> causing ebuilds to bail out? How can you disable such checks per-ebuild |
14 |
> (AFAIK - you can't) to not annoy users with QA notices and breakage one can |
15 |
> do nothing about anyway? |
16 |
> |
17 |
> As Flameeyes pointed out, dodoc/dohtml is also used in eclasses. This can |
18 |
> break many ebuilds. Users will report duplicate bugs because they will not |
19 |
> realize that it's the eclass causing the failure, not the ebuild. Again, |
20 |
> what's the point? How will it work with FEATURES="nodoc"? Why should an |
21 |
> ebuild ever fail just because some doc file is missing or got renamed or |
22 |
> whatever? |
23 |
> |
24 |
> |
25 |
|
26 |
echo "dodoc: ${x} does not exist" 1>&2 |
27 |
|
28 |
This should not be showed to users then because we know this to happen |
29 |
and do not want to do anything about it. Also try to remember that not |
30 |
every package is done following the GNU conventions (Java). It can |
31 |
happen that the doc files are for example renamed and dieing will help |
32 |
catch this. dodoc/dohtml dieing will not have any affect on |
33 |
FEATURES="nodoc". |
34 |
|
35 |
How about making it possible to tell dodoc/dohtml not to die as an |
36 |
argument/variable/whatever? We could also have new diedodoc functions, |
37 |
but it would be nice to keep this as simple as possible. |
38 |
|
39 |
Regards, |
40 |
Petteri |