1 |
On Sun, Mar 2, 2014 at 1:09 PM, Jeroen Roovers <jer@g.o> wrote: |
2 |
> Honestly, setting up a tracker and blocking it with bugs about packages |
3 |
> which someones-sub-SLOT-checking-script has vetted to be involved could |
4 |
> be done in less than a day (for the hundred or so packages that depend |
5 |
> on dev-libs/libgcrypt). It doesn't need some QA team to study in depth |
6 |
> -- it needs a couple of volunteers to do the checks and file the bug |
7 |
> reports. |
8 |
|
9 |
Honestly, it would be best to coordinate this with QA. QA isn't just |
10 |
about taking away commit rights in punishment for your sins. Scripts |
11 |
to help make devs aware of potential improvements should be a part of |
12 |
QA as well. |
13 |
|
14 |
And that isn't saying that you have to be on the QA team in order to |
15 |
do this either. If this sort of thing interests anybody, just ping |
16 |
them. I know QA is interested in improving the use of slot operator |
17 |
dependencies, because they've said as much on their irc channel... |
18 |
|
19 |
My script for spotting opportunities to use slot op deps is still out |
20 |
there. Of course, I wouldn't just "fix" any ebuild that comes up as |
21 |
there are situations where it isn't appropriate. What would probably |
22 |
make sense is for QA to define a var for ebuilds to set when they've |
23 |
been screened and found to be false positives, and then suppress them |
24 |
from the output. |
25 |
|
26 |
I'm all for nudging maintainers to add slot op deps when they make |
27 |
sense, or allowing others to do the work for them as long as they |
28 |
exercise reasonable care. Ebuilds don't belong to maintainers, but we |
29 |
should work with them whenever possible so that they're in a position |
30 |
to spot issues and deal with them. |
31 |
|
32 |
Rich |