Gentoo Websites Logo
Go to: Gentoo Home Documentation Forums Lists Bugs Planet Store Wiki Get Gentoo!
Bug 190746 - permission resetting of unpacked content
Summary: permission resetting of unpacked content
Status: RESOLVED OBSOLETE
Alias: None
Product: Gentoo Linux
Classification: Unclassified
Component: New packages (show other bugs)
Hardware: AMD64 Linux
: Normal enhancement
Assignee: Sven Wegener
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2007-08-30 05:21 UTC by Mister Woody
Modified: 2012-04-14 17:02 UTC (History)
3 users (show)

See Also:
Package list:
Runtime testing required: ---


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Mister Woody 2007-08-30 05:21:23 UTC
paludis -i screen  ( I suppose also emerge screen) gives me:

>>> Unpacking screen-4.0.3_p20070403.tar.gz to
/var/tmp/paludis/app-misc/screen-4.0.3_p20070403/work
tar zxf /usr/portage/distfiles/screen-4.0.3_p20070403.tar.gz --no-same-owner
 * Applying screen-4.0.3_p20070403-map.patch ...                         [ ok ]
 * Applying 4.0.2-no-utempter.patch ...                                  [ ok ]
 * Applying 4.0.2-no-libelf.patch ...                                    [ ok ]
 * Running autoconf ...     

 * Failed Running autoconf !
 *
 * Include in your bugreport the contents of:
 *
 * /var/tmp/paludis/app-misc/screen-4.0.3_p20070403/temp//autoconf-5510.out


---------------------------------------------------


cat /var/tmp/paludis/app-misc/screen-4.0.3_p20070403/temp//autoconf-5510.out


autom4te-2.61: cannot open configure: Permission denied
Comment 1 Jakub Moc (RETIRED) gentoo-dev 2007-08-30 06:55:37 UTC
(In reply to comment #0)
> paludis -i screen  ( I suppose also emerge screen) gives me:

Not really, emerge =app-misc/screen-4.0.3_p20070403 works perfectly fine for me.

Comment 2 Ciaran McCreesh 2007-08-30 14:06:37 UTC
(In reply to comment #1)
> (In reply to comment #0)
> > paludis -i screen  ( I suppose also emerge screen) gives me:
> 
> Not really, emerge =app-misc/screen-4.0.3_p20070403 works perfectly fine for
> me.

That sort of comment isn't helpful. Working or not working for one person is not indicative either way on whether it is package manager related.
Comment 3 Mister Woody 2007-08-30 14:12:51 UTC
I should have added that I have been using amd64
Comment 4 Jakub Moc (RETIRED) gentoo-dev 2007-08-30 14:23:45 UTC
(In reply to comment #2)
> That sort of comment isn't helpful. Working or not working for one person is
> not indicative either way on whether it is package manager related.

With this kind of 'logic', noone could ever comment in a helpful way unless that person is suffering from split personality.

As noted above, it works fine with portage, if fails w/ paludis and it fails w/ pkgcore. So, it totally is package manager related.
Comment 5 Mister Woody 2007-08-30 14:35:46 UTC
it works fine for portage also here. So it is paludis related. I will close this bug. 
Thanks!
Comment 6 Jakub Moc (RETIRED) gentoo-dev 2007-08-30 14:38:44 UTC
Well, closing the bug doesn't exactly help to paludis/pkgcore users. :)
Comment 7 David Leverton 2007-08-30 15:39:56 UTC
The issue is that the tarball includes configure without write permissions; portage resets the permissions after unpacking, but paludis and pkgcore don't.  This is fixed in paludis r3455.  CCing ferringb in case he wants to do the same in pkgcore.

I'll let swegener decide whether or not to add a workaround to the ebuild for now.
Comment 8 Ciaran McCreesh 2007-08-30 20:28:47 UTC
(In reply to comment #4)
> (In reply to comment #2)
> > That sort of comment isn't helpful. Working or not working for one person is
> > not indicative either way on whether it is package manager related.
> 
> With this kind of 'logic', noone could ever comment in a helpful way unless
> that person is suffering from split personality.

No, Jakub. One could establish it by using identical configurations on an identical box with multiple package managers. If only one variable is changed, *then* it's fairly likely to be indicative; if lots of variables are changed, it could be lots of things.
Comment 9 Brian Harring (RETIRED) gentoo-dev 2007-09-01 00:44:18 UTC
This needs to be updated in pms.
Comment 10 Brian Harring (RETIRED) gentoo-dev 2007-09-01 01:37:04 UTC
(In reply to comment #9)
> This needs to be updated in pms.

Actually, on second thought- this really ought to be discussed on the ml imo; there are quite a few distfiles that do some fun things (trying to change the perms of the actual WORKDIR for example via ./ entries); strikes me establishing a base req of what the manager is actually expected to do (what is/is not sane).

I'll likely add a hack for this, but this ought to be nailed down, and actually documented (instead of just parroting what portage is doing as the last few cases have triggered).
Comment 11 Brian Harring (RETIRED) gentoo-dev 2008-03-06 07:37:44 UTC
What's the status of this bug in re: to pms docs?
Comment 12 Ciaran McCreesh 2008-03-06 09:18:06 UTC
(In reply to comment #11)
> What's the status of this bug in re: to pms docs?

It's made it as far as my bit of paper with "things to review or rewrite" on it.
Comment 13 Brian Harring (RETIRED) gentoo-dev 2009-10-19 05:44:28 UTC
19 months later... should be doing something to close this out if it's still relevant pms wise.
Comment 14 Ciaran McCreesh 2009-10-19 09:06:21 UTC
Given that we've only ever seen a single affected package... I'm inclined to just leave it until more crop up. Doesn't seem to be something that happens much.
Comment 15 Ulrich Müller gentoo-dev 2012-04-14 17:02:25 UTC
(In reply to comment #7)
> The issue is that the tarball includes configure without write permissions;
> portage resets the permissions after unpacking, but paludis and pkgcore
> don't.  This is fixed in paludis r3455.  CCing ferringb in case he wants to
> do the same in pkgcore.

$ tar tvf /usr/portage/distfiles/screen-4.0.3.tar.gz | grep configure
-rwxr-xr-x mls/suse     251351 2003-12-05 14:46 screen-4.0.3/configure
-rw-r--r-- mls/suse      30190 2003-06-03 13:58 screen-4.0.3/configure.in

Looks like the problem existed only with the tarball of the CVS snapshot, whose ebuild has been removed two years ago.

PMS wise, I believe that the issue has been clarified by this commit: <http://git.overlays.gentoo.org/gitweb/?p=proj/pms.git;a=commit;h=a862eae3a329945266879ebe0c3142c3367f5c3d>

Closing.