1 |
On Thu, 2006-03-02 at 20:49 +0000, Ciaran McCreesh wrote: |
2 |
> Now, you've heard that dropping keywords is bad. But you have a clever |
3 |
> idea, and make the dep alsa? ( !sparc? ( alsa libraries ) ). This gets |
4 |
> past repoman just fine. |
5 |
> |
6 |
|
7 |
<STOP> As any arch can tell you, that's never stopped me - *IF* you do |
8 |
it correctly, ie comment out the existing keywords, add the keywords |
9 |
that the package can support, and file a bug against the arch's that you |
10 |
had to drop explaining the need for re-keywording because of a new dep |
11 |
that they don't yet support. AFAIK that's the correct way to do it - and |
12 |
I believe that pretty strongly since at this point there isn't a single |
13 |
arch that wouldn't have filed a grievance against me otherwise. Sure, |
14 |
those bugs may stay open for months and months and months because ia64 |
15 |
doesn't have the resources to devote (which is understood), but at that |
16 |
point you have 2 references for users, the ebuild with the commented out |
17 |
line, the ChangeLog, and the bug. |
18 |
|
19 |
Is this not how its supposed to be done? Because if it is, maybe those |
20 |
insistant on the !arch method should be pointed to that and leave it up |
21 |
to the arch's to make the decision of whether to keyword or disable |
22 |
specific support. Devs acting on behalf of a herd shouldn't be making |
23 |
these kind of arch decisions, but instead leaving it up to devs acting |
24 |
with their arch hats on. Maybe the two meet under the same roof |
25 |
sometimes, but more times than not they don't. Yep, that's a lot more |
26 |
work and effort and pain - but afaik it follows good qa methods. |
27 |
|
28 |
And despite the length of this message, I haven't spoken on one side of |
29 |
the fence or the other I think...sweet :) |
30 |
|
31 |
~mcummings |