1 |
M. J. Everitt posted on Sun, 13 Aug 2017 11:18:09 +0100 as excerpted: |
2 |
|
3 |
> On 13/08/17 11:11, Michael Orlitzky wrote: |
4 |
>> On 08/12/2017 10:52 PM, Duncan wrote: |
5 |
>>> How so? Are you arguing that deciding to system-wide switch to/from |
6 |
>>> pulseaudio, systemd, or gstreamer is nonsense? |
7 |
>>> |
8 |
>> The meaning of any one USE flag varies widely across packages. I could |
9 |
>> never say "I want to enable USE=gstreamer" for every package in the |
10 |
>> tree, because I have no idea what it does for most of them. Setting |
11 |
>> USE=whatever globally essentially means "make random changes to my |
12 |
>> system" -- hence my wording. |
13 |
>> |
14 |
>> The meaning of a USE flag is per-package, so per-package is the only |
15 |
>> meaningful way to set them. |
16 |
>> |
17 |
> Which is why we have GLOBAL use flags and LOCAL use flags, right?! |
18 |
> |
19 |
> I'm not sure I'm actually reading this discussion right now?! and I'm |
20 |
> *not* a dev ... |
21 |
|
22 |
Even then... given the descriptions of most flags, regardless of global |
23 |
or local, without reading the sources (assuming one /can/ read them) it's |
24 |
generally a WAG what a flag actually does on an individual package. The |
25 |
information simply isn't there, except in the sources, and few people |
26 |
care enough about what a flag does to actually go the work of reading |
27 |
/all/ affected sources. |
28 |
|
29 |
[TL;DR stop here] |
30 |
|
31 |
FWIW, here I set USE="-* ..." in make.conf[1], so I can't /not/ have a |
32 |
global policy. If the flag's in make.conf, global policy says it's on. |
33 |
If it's not, global policy says it's off. There's no longer an "on if |
34 |
someone else defaults it that way, subject to change without pre- |
35 |
notification" state. After a few years chasing down reasons for changes |
36 |
to see if I agreed or not, I simply found it simpler to ensure I was the |
37 |
only one toggling USE flags, in which case I'd know exactly why I did it. |
38 |
|
39 |
And it works surprisingly well, too, particularly when package.use is |
40 |
multiple files, each named after a flag, "0-flag" for negating exceptions |
41 |
to an on global policy, simply "flag" for "posating" exceptions to an off |
42 |
global policy. With dates and comments for each exception and all the |
43 |
exceptions in one place (it's easy to equery uses the settings for a |
44 |
package, not so easy to find all the exceptions to a global policy |
45 |
setting and why, unless the exceptions are all in one place organized by |
46 |
flag name), management is as simple as it's going to get. |
47 |
|
48 |
-- |
49 |
Duncan - List replies preferred. No HTML msgs. |
50 |
"Every nonfree program has a lord, a master -- |
51 |
and if you use the program, he is your master." Richard Stallman |