1 |
On wto, 2017-05-30 at 04:30 +1200, Kent Fredric wrote: |
2 |
> On Mon, 29 May 2017 17:33:13 +0200 |
3 |
> Michał Górny <mgorny@g.o> wrote: |
4 |
> |
5 |
> > Automatically solving USE constraints solve all three fore-mentioned |
6 |
> > issues with REQUIRED_USE. By default, no user intervention is required |
7 |
> > to solve USE constraints and package.use needs to be modified only to |
8 |
> > enforce a non-standard solutons |
9 |
> |
10 |
> Overall I like the proposal, but one question: How do you envisage |
11 |
> automatic use-constraints interacting with --newuse? |
12 |
> |
13 |
> I have this feeling that "automatically enabled" flags could somehow |
14 |
> have an ephemeral nature, where a flag would be enabled at build time, |
15 |
> and then later a subsequent change in the graph toggles the flag off, |
16 |
> creating a potentially undesirable rebuild. |
17 |
> |
18 |
> I feel I might be imagining a problem because I might have a wire |
19 |
> crossed somewhere, so some sort of confirmation that I'm the insane one |
20 |
> and this can't happen would be reassuring :) |
21 |
|
22 |
I might be missing something but I don't think there would be any |
23 |
problems that we don't have right now. To the contrary, this proposal |
24 |
specifically reduces the amount of USE flag changes possible as flags |
25 |
can be more strongly bound to valid sets only. |
26 |
|
27 |
That said, the code running --newuse/--changed-use would probably need |
28 |
to account for the constraints, i.e. include them in calculating |
29 |
the effective set of USE flags. However, it's not much different from |
30 |
handling use.force/use.mask, and all the metadata is in place, so it |
31 |
shouldn't impact performance noticeably. |
32 |
|
33 |
-- |
34 |
Best regards, |
35 |
Michał Górny |