1 |
On Sat, Jun 23, 2012 at 10:24 AM, Ciaran McCreesh |
2 |
<ciaran.mccreesh@××××××××××.com> wrote: |
3 |
> On Sat, 23 Jun 2012 10:16:42 -0700 |
4 |
> Alec Warner <antarus@g.o> wrote: |
5 |
>> I don't think the documentation forbids what these developers are |
6 |
>> doing. |
7 |
> |
8 |
> http://www.gentoo.org/proj/en/devrel/handbook/handbook.xml?part=2&chap=1 |
9 |
> |
10 |
> "This means that counting goes as follows: 1.0 (initial version), |
11 |
> 1.0-r1, 1.0-r2, etc." |
12 |
> |
13 |
> It's not illegal, but it's also not in line with how versions and slots |
14 |
> have interacted up until now. |
15 |
|
16 |
I agree and I sympathize with your position. |
17 |
|
18 |
> |
19 |
>> I think you implemented a nice heuristic for your users in your |
20 |
>> resolver that used to work because slots had a typical set of users |
21 |
>> cases and the heuristic performed well in those cases. |
22 |
>> |
23 |
>> Now people are occasionally using slots in a different way and your |
24 |
>> heuristic penalizes those cases. That sucks, but you might have to |
25 |
>> actually change your resolver because I don't think 'funky-slots' |
26 |
>> properties is going to garner much adoption. |
27 |
> |
28 |
> You mean, instead of implementing trivial marking, which takes |
29 |
> developers a few seconds, you want to screw over users? I think that |
30 |
> says a lot about Gentoo's attitude... |
31 |
|
32 |
I don't think portage has the behavior that paludis does, so most |
33 |
users are not likely to experience this particular problem. You know |
34 |
as well as I that the marking isn't necessarily trivial. Its another |
35 |
thing we have to document and train people to use. I don't think users |
36 |
get 'screwed over' either. |
37 |
|
38 |
It could be that instead of Gentoo tagging a bunch of ebuilds, you |
39 |
just change your resolver heuristic a bit. |
40 |
|
41 |
-A |
42 |
|
43 |
> |
44 |
> -- |
45 |
> Ciaran McCreesh |