1 |
On Tue, Jun 23, 2015 at 08:05:35AM +0200, Michał Górny wrote: |
2 |
> Dnia 2015-06-23, o godz. 01:48:58 |
3 |
> Alexandre Rostovtsev <tetromino@g.o> napisał(a): |
4 |
> |
5 |
> > On Tue, 2015-06-23 at 07:06 +0200, Michał Górny wrote: |
6 |
> > > Dnia 2015-06-23, o godz. 01:23:13 |
7 |
> > > Jason Zaman <jason@×××××××××.com> napisał(a): |
8 |
> > > |
9 |
> > > > On Mon, Jun 22, 2015 at 10:55:45PM +0200, Michał Górny wrote: |
10 |
> > > > > Dnia 2015-06-22, o godz. 16:38:30 |
11 |
> > > > > Jason Zaman <perfinion@g.o> napisał(a): |
12 |
> > > > > |
13 |
> > > > > > Hi all, |
14 |
> > > > > > |
15 |
> > > > > > I want to add STRONGSWAN_PLUGINS to USE_EXPAND. This is related |
16 |
> > > > > > to bug |
17 |
> > > > > > https://bugs.gentoo.org/show_bug.cgi?id=504942 |
18 |
> > > > > > "net-misc/strongswan missing USE flags for plugins" |
19 |
> > > > > > |
20 |
> > > > > > Patrick said to fix it myself, the ebuilds with the plugins |
21 |
> > > > > > have |
22 |
> > > > > > been in the tree for a fair while now and only the USE_EXPAND |
23 |
> > > > > > is |
24 |
> > > > > > missing. |
25 |
> > > > > > |
26 |
> > > > > > If there are no objections, I will commit the following patch |
27 |
> > > > > > on friday: |
28 |
> > > > > |
29 |
> > > > > USE_EXPAND is global by its nature. USE flags used by a single |
30 |
> > > > > package |
31 |
> > > > > are not appropriate for making global. |
32 |
> > > > |
33 |
> > > > Since when? There are so many things in USE_EXPAND that are only |
34 |
> > > > for a |
35 |
> > > > single package. |
36 |
> > > |
37 |
> > > Not sure when. But the policies about USE flags were in the devmanual |
38 |
> > > long before I came here. |
39 |
> > > |
40 |
> > > > APACHE2_MODULES, NGINX_MODULES_HTTP, COLLECTD_PLUGINS, QEMU_*, etc. |
41 |
> > > > Easily half of the things in USE_EXPAND are for only a single |
42 |
> > > > package. |
43 |
> > > |
44 |
> > > Past screwups don't justify future screwups. |
45 |
> > |
46 |
> > I don't see it as a screwup personally. |
47 |
> > |
48 |
> > It is often useful to have different namespaces for the few general |
49 |
> > flags that control a package's general features and the >9000 flags |
50 |
> > that enable/disable >9000 highly specific plugins. |
51 |
> > |
52 |
> > And if violation of the rules is useful, perhaps the rules are wrong? |
53 |
> |
54 |
> Then perhaps you go and fix the rules instead of ignoring them? |
55 |
> |
56 |
> But please also remember to provide the ability to describe those flags |
57 |
> per-package rather than globally, like they are done now. Would be good |
58 |
> to also avoid declaring them globally, like having >50 groups listed in |
59 |
> USE_EXPAND. |
60 |
> |
61 |
> When you're done with that, and get all package managers to support it |
62 |
> in a reasonably long stable version, then we can discuss about changing |
63 |
> the rules. |
64 |
|
65 |
I dont quite follow, the useflags are described in the packages |
66 |
metadata.xml, are not conflicting and are supported by everything. |
67 |
|
68 |
Is there a rule that everything that is USE_EXPAND'd is a global |
69 |
useflag? I see no such rule and I dont understand why it would even make |
70 |
sense. USE_EXPAND arnt useflags, it is just a variable to make useflags |
71 |
look nicer, they shouldnt all have to be global. What if one and only |
72 |
one package supports a specific video_card? Should that use-flag be made |
73 |
global? |
74 |
|
75 |
Adding use_expand's for foo_plugins makes things a lot easier to manage |
76 |
and understand tho. The only possible downside I can see is that there |
77 |
are a lot of entries in it, but we have a lot of useflags so thats hardly |
78 |
an issue. |
79 |
|
80 |
-- Jason |