1 |
On Sat, 22 Jul 2006 16:52:38 -0700 |
2 |
Zac Medico <zmedico@g.o> wrote: |
3 |
|
4 |
> -----BEGIN PGP SIGNED MESSAGE----- |
5 |
> Hash: SHA1 |
6 |
> |
7 |
> Mike Kelly wrote: |
8 |
> |
9 |
> > In my case, I feel this functionality would be very useful as it |
10 |
> > allows for me to integrate my GLEP 27 implementation into portage |
11 |
> > without portage needing to worry about my implementation specifics, |
12 |
> > which may well change in later versions. I am sure there are other |
13 |
> > practical ways in which these hooks could be used. |
14 |
> > |
15 |
> > Patch at: |
16 |
> > http://soc.gentoo.org/viewcvs.py/*checkout*/glep0027/patches/portage-hooks.patch |
17 |
> > |
18 |
> > See bug# 141215. |
19 |
> > |
20 |
> |
21 |
> I like the idea. My main concern is that, like /etc/portage/bashrc, |
22 |
> this creates lots of potential for foreign code to interfere with the |
23 |
> internal workings of the ebuild environment. At a minimum, I think |
24 |
> there should be something in the `emerge --info` output that |
25 |
> indicates whether or not any phase hooks exist. |
26 |
> |
27 |
> Traditionally, configurable files that affect portage behavior go |
28 |
> mostly in /etc/portage. I see that your intention is to |
29 |
> make /var/libexec/portage/hooks/ a location for third-party packages |
30 |
> to install hooks, so it makes sense to keep those separate from hooks |
31 |
> that the user might install. I know that portage-utils currently |
32 |
> installs a post-sync hook in /etc/portage/postsync.d/q-reinitialize, |
33 |
> so there is some inconsistency there. We need to develop a |
34 |
> consistent policy for appropriate locations of files like these. |
35 |
|
36 |
I still think that those hooks should go into the tree rather than |
37 |
being installed by packages. That's mainly so they get increased |
38 |
visibility and gives us (us being Gentoo, not just Portage) more |
39 |
access and control over them, like we have with eclasses. |
40 |
|
41 |
This also moves responsibility from hook authors to pkgmanager authors |
42 |
(the package mamanger has to support the hook format in the tree, not |
43 |
the hook has to support (all of) the specific package manager hook |
44 |
formats). I know Mike has taken care of Paludis support already, but |
45 |
look at a larger scale: |
46 |
a) n hook authors supporting m package manager interfaces: O(n*m) |
47 |
b) n hook authors and m package managers supoprting one repository |
48 |
interface: O(n+m) |
49 |
Also with a) you have to play what I call the "catch up game", e.g. a |
50 |
new package manager gets out and all hooks have to account for a new |
51 |
interface (even if it's just a new path). |
52 |
Of course a) has the "drawback" that it requires a solid spec of the |
53 |
interface, but that's something we want anyway. |
54 |
|
55 |
However, *if* they are going to be installed by packages they should go |
56 |
either into /etc/portage/foobar or /usr/lib/portage/foobar (like any |
57 |
3rd party elog or cache modules), foobar being something we'd have to |
58 |
decide on. |
59 |
The suggested /var/libexec/portage has several issues: |
60 |
- /var/libexec is AFAIK not defined by FHS |
61 |
- executable data doesn't belong in /var unless it's application data |
62 |
which isn't the case here |
63 |
- libexec generally sucks (even in /usr), especially as we already |
64 |
use /usr/lib for what generally goes there (support scripts) |
65 |
|
66 |
Independent on the location I'd like to see the ability to blacklist |
67 |
individual hooks for testing, troubleshooting, debugging and probably a |
68 |
few other reasons. And no, this should not be done with $FEATURES ;) |
69 |
|
70 |
Marius |
71 |
|
72 |
-- |
73 |
Public Key at http://www.genone.de/info/gpg-key.pub |
74 |
|
75 |
In the beginning, there was nothing. And God said, 'Let there be |
76 |
Light.' And there was still nothing, but you could see a bit better. |