1 |
On Sat, Jun 23, 2012 at 9:37 PM, Ciaran McCreesh |
2 |
<ciaran.mccreesh@××××××××××.com> wrote: |
3 |
> On Sat, 23 Jun 2012 21:35:47 +0300 |
4 |
> Alex Alexander <wired@g.o> wrote: |
5 |
>> > The package mangler does not know that 1.1-r300 is not a "better" |
6 |
>> > version than 1.1-r200, or that 1.2-r200 is not a "better" version |
7 |
>> > than 1.1-r300. Indicating packages where this kind of strangeness |
8 |
>> > happens allows manglers to know that things that are usually true |
9 |
>> > about the relationship between slots and versions no longer hold, |
10 |
>> > and that in these specific cases it should consider slots to be |
11 |
>> > heavily independent. |
12 |
>> |
13 |
>> You already have this info, it's called a "slot dependency". |
14 |
> |
15 |
> It's not a property of individual packages that happen to depend upon |
16 |
> the problematic package. The property holds or not for a package |
17 |
> regardless of whether anything depends upon it. |
18 |
|
19 |
They are part of the deal. |
20 |
|
21 |
If your package has reverse deps, you don't want to update it before |
22 |
figuring out it's reverse dependencies anyway, you never know what |
23 |
slot/version restrictions you're going to get. |
24 |
|
25 |
If it is a package without reverse dependencies, updating to the most |
26 |
recent slot and/or version should be expected unless the user has the |
27 |
slot defined in the world file. |
28 |
|
29 |
-- |
30 |
Alex |