1 |
On Friday 21 September 2007, Chris Gianelloni wrote: |
2 |
> On Thu, 2007-09-20 at 20:52 -0400, Mike Frysinger wrote: |
3 |
> > On Thursday 20 September 2007, Chris Gianelloni wrote: |
4 |
> > > Also, my plan for this would be add the copying of the .bash* files |
5 |
> > > from /etc/skel if and only if USE=build *and* the files do not already |
6 |
> > > exist, done during pkg_preinst/pkg_postinst somewhere. This will pull |
7 |
> > > it into the stage1 tarball, making it available to everyone on all |
8 |
> > > further stages, it keeps it from being done every time someone emerges |
9 |
> > > bash, it doesn't stomp on existing files, and it doesn't show up in VDB |
10 |
> > > linked to the bash package. Is this acceptable? |
11 |
> > |
12 |
> > no, bash is not a special case |
13 |
> > |
14 |
> > why not add an `emerge --config baselayout` to the end of stage3 and i'll |
15 |
> > do the /root/ default setup in the baselayout ebuild |
16 |
> |
17 |
> I'd rather not do anything that isn't accomplished automatically by the |
18 |
> ebuild when it is being merged. If this really is so objectionable, I'd |
19 |
> just assume WONTFIX the request and move on with it. |
20 |
|
21 |
as outlined already, it's objectionable in the sense that once a user has |
22 |
entered the system, nothing can be assumed in the automatic case. by the |
23 |
user running `emerge --config baselayout`, we can assume that the behavior it |
24 |
exhibits (for example, setting up files in /root that dont exist |
25 |
from /etc/skel/) is expected. |
26 |
|
27 |
as also explained, tying into USE=bootstrap and USE=build is not correct as |
28 |
that does not represent everything in stage3 and there is no way to detect |
29 |
the stage2->stage3 state vs a normal emerge. |
30 |
|
31 |
the compromise is simple: catalyst runs --config at the end of stage3 for |
32 |
appropriate packages, but as to what those things actually do is left in the |
33 |
ebuilds. |
34 |
-mike |