1 |
On Wed, Aug 30, 2017 at 7:36 AM, Michał Górny <mgorny@g.o> wrote: |
2 |
> W dniu śro, 30.08.2017 o godzinie 14∶08 +0200, użytkownik Kristian |
3 |
> Fiskerstrand napisał: |
4 |
>> On 08/30/2017 01:59 PM, Ulrich Mueller wrote: |
5 |
>> > > > > > > On Tue, 29 Aug 2017, Ulrich Mueller wrote: |
6 |
>> > > |
7 |
>> > > I request the Council to approve a PMS change, namely to ban empty |
8 |
>> > > dependency groups like "|| ( )" or "foo? ( )". |
9 |
>> > |
10 |
>> > I retract this. It appears to be controversial, which is bad for a |
11 |
>> > retroactive change. (And IMHO it is an obscure corner case anyway.) |
12 |
>> > |
13 |
>> |
14 |
>> It might still be a good idea to discuss the matter, although I agree it |
15 |
>> likely is better to discuss it more broadly before it reaches the |
16 |
>> council in such cases. |
17 |
>> |
18 |
>> If anything it sounds like PMS might require some updating to be more |
19 |
>> explicit on the handling of such cases if there is uncertainty on how to |
20 |
>> interpret it. |
21 |
>> |
22 |
>> > Let's make Portage comply with the spec instead. |
23 |
>> |
24 |
>> That seems like a good idea in any case :) |
25 |
>> |
26 |
> |
27 |
> I see no reason to make Portage less strict than it is now. There is no |
28 |
> valid use case for this, and the PMS behavior is just plain stupid. |
29 |
> Reverting the change will not bring any clear gain, and will only |
30 |
> surprise people who actually hit the case. |
31 |
> |
32 |
|
33 |
++ |
34 |
|
35 |
Measure twice, cut once. Figure out what the spec should be before we |
36 |
start changing things, unless we're confident that the direction we're |
37 |
going to head down is actually an improvement. The fact that getting |
38 |
it right is harder than was originally thought is just that much more |
39 |
reason to not mess with the code yet if nothing is actually breaking. |
40 |
|
41 |
-- |
42 |
Rich |