On Tue, 2004-03-23 at 18:54, Paul Smith wrote:
> %% david@... writes:
> 
>   d> On Tue, Mar 23, 2004 at 01:57:06PM -0600, Joseph Booker wrote:
> 
>   >> for merging the packages..........$ROOT sounds like to me (and I
>   >> havn't checked) that when an emerge is done, when it 'merges'
>   >> /..../image with /, it insteads merges it with $ROOT, for example,
>   >> if $ROOT was /usr, then all the files installed in /etc would to
>   >> got /usr/etc, all in /usr/bin goto /usr/usr/bin, etc.
> 
>   d> I have use $ROOT quite extensively.  I can vouch for it's working
>   d> in this fashion, the only time it doesn't produce expected results
>   d> is specific checks in ebuilds.  Some ebuilds will check for a
>   d> library(in the actual ebuild, not a configure script, ex: perl
>   d> checks for libperl) and the check will be done on / not $ROOT So if
>   d> the file/library required exists on $ROOT but not / the build
>   d> fails.
> 
> Either I'm doing something wrong, or I'm still not understanding, or
> there are problems in $ROOT handling.
> 
> 
> I just tried a basic setup with a vanilla Portage install without any of
> our local changes (from Portage CVS circa 2003/03/19).
> 
> When I run "ROOT=/tmp/ptest emerge someapp" I get these errors:
> 
>   >>> /tmp/ptest/tmp doesn't exist, creating it...
>   >>> /tmp/ptest/var/tmp doesn't exist, creating it...
>   Traceback (most recent call last):
>     File "/tmp/ptest/usr/lib/portage/bin/emerge", line 14, in ?
>       import portage
>     File "/tmp/ptest/usr/lib/portage/pym/portage.py", line 6320, in ?
>       os.makedirs(cachedir,0755)
>     File "/usr/lib/python2.2/os.py", line 203, in makedirs
>       mkdir(name, mode)
>   OSError: [Errno 13] Permission denied: '/var/cache/edb'
> 
> 
> Looking the code, it's looking for _BOTH_ /var/cache/edb _AND_
> $ROOT/var/cache/edb.

I would think that is correct, to test for both DEPEND _AND_ RDEPEND. 
The stuff in DEPEND would need to be in your REAL /, whereas RDEPEND
would need to be in $ROOT/.

> I don't want to have to have the former, if $ROOT is set.  In fact in
> our environment it won't exist.  Is there a reason we need to test for
> both?

Definitely.

> Anyway, I'll continue with this and report problems as I find them.
> Is this better discussed on the gentoo-portage-dev list?  Or should I
> just file bugs in Gentoo Bugzilla without discussion?  Or...?

Well, in this case discussion has saved an unnecessary bug being filed,
so I would say that it is a good thing.  I'm pretty interested in what
you're doing and saying, and I think quite a few others are judging by
the other responses.  Also, while this is mostly portage related, I
think it also would be useful for catalyst and even for people doing
full-fledged virtual services.  I think it should stay on -dev provided
it doesn't get too specific to portage alone.

-- 
Chris Gianelloni
Developer, Gentoo Linux
Games Team

Is your power animal a penguin?