1 |
On Tue, Sep 13, 2005 at 11:51:18AM -0500, Grant Goodyear wrote: |
2 |
> Grant Goodyear wrote: [Tue Sep 13 2005, 11:40:43AM CDT] |
3 |
> > I'm not sure that's entirely correct. I seem to remember at least one |
4 |
> > devrel dev stating that when it comes to devs who violate technical |
5 |
> > policies (not using repoman, repeatedly breaking sections of the tree, |
6 |
> > etcetera) that enforcement should be left up to the appropriate |
7 |
> > managers, not devrel. The argument was that devrel devs are often not |
8 |
> > experts in the technical aspects, so it's hard for them to adjudicate |
9 |
> > effectively. |
10 |
> |
11 |
> I should also mention that I'm not advocating this interpretation. I'd |
12 |
> much prefer that devrel's scope encompass such technical issues. |
13 |
|
14 |
I'd prefer the QA project/herd handle this. |
15 |
|
16 |
In my opinion, devrel should deal in developer pissing matches |
17 |
(preferably kicking both parties in the head for fighting), incoming |
18 |
devs, outgoing devs, and carrying out punitive measures. |
19 |
|
20 |
QA involves a helluva lot more then just reacting when people complain |
21 |
that XYZ is screwing up the tree; proper QA involves actually |
22 |
identifying xyz is screwing up the tree rather then a reactive |
23 |
approach. |
24 |
|
25 |
Essentially, QA requires people actively auditing the tree, deps, and |
26 |
nudging devs to stop screwing things up, preferably with advice on how |
27 |
to avoid screwing up. This involves a good chunk of work, and for the |
28 |
work to actually go anywhere, there needs to be backing of some sort. |
29 |
|
30 |
QA has never had true backing beyond (essentially) whining to devrel |
31 |
that xyz is breaking stuff. It's not particularly surprising that |
32 |
they haven't been incredibly effective, considering that fact. |
33 |
|
34 |
Yes, Mr_bones_ will rightfully tear your ass if you keep breaking |
35 |
things, but ultimately it's just nagging, if he wants anything done he |
36 |
has to present the case to devrel, who may or may not do something. |
37 |
|
38 |
This setup I view as (bluntly) broke; devrel isn't tracking what's |
39 |
going on in the tree, Michael is, further he's tracking who screws |
40 |
up and who doesn't on a regular basis due to his scans. He knows who |
41 |
has been naughty or nice, essentially :) |
42 |
|
43 |
Dunno, my two cents. Not much for QA being under the auspices of |
44 |
devrel for the reasons above, but also keeping things seperated, and |
45 |
avoiding more cabal bitching. |
46 |
|
47 |
Not meaning this to be a slap in devrel's direction mind you; question |
48 |
of area of focus. They deal in hauling in devs, dealing with idiot |
49 |
devs, and chucking awol devs; I really don't see how QA falls under |
50 |
them beyond potentially the punitive aspect of QA having someone's cvs |
51 |
turned off for continually screwing up (willingly or otherwise). |
52 |
~harring |