Gentoo Websites Logo
Go to: Gentoo Home Documentation Forums Lists Bugs Planet Store Wiki Get Gentoo!
Bug 906532 - net-misc/openntpd-6.2_p3-r2: USE libressl was removed but revision was not incremented
Summary: net-misc/openntpd-6.2_p3-r2: USE libressl was removed but revision was not in...
Status: RESOLVED INVALID
Alias: None
Product: Gentoo Linux
Classification: Unclassified
Component: Current packages (show other bugs)
Hardware: All Linux
: Normal normal
Assignee: Paul B. Henson
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2023-05-16 08:44 UTC by Jernej Jakob
Modified: 2023-05-18 14:05 UTC (History)
2 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 Jernej Jakob 2023-05-16 08:44:36 UTC
In this commit the ebuild was modified to remove the libressl use flag
https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=beea04c481c315ecfe368a9fb70c0eddf5ab6805
but the package revision was not incremented after that.
This causes installation from binary package that was created before its removal to fail because of non-matching use flags.

Reproducible: Always

Steps to Reproduce:
1. binpkg host creates the package before commit beea04c4 with libressl use flag still present (whether it's enabled or not, doesn't matter)
2. binpkg consumer tries to install the package after beea04c4 which causes it to detect non-matching use flags and fall back to a source build (if so configured)
3. a simple 'emerge -1 net-misc/openntpd' on the binpkg host fixes the issue
Comment 1 Michał Górny archtester Gentoo Infrastructure gentoo-dev Security 2023-05-16 15:29:52 UTC
I'd say this is INVALID.  It doesn't seem justified to force rebuild on everyone to workaround what is essentially a limitation in your binary package builder.
Comment 2 Jernej Jakob 2023-05-16 20:43:09 UTC
(In reply to Michał Górny from comment #1)
> I'd say this is INVALID.  It doesn't seem justified to force rebuild on
> everyone to workaround what is essentially a limitation in your binary
> package builder.

Then should this be a portage bug/improvement instead? If the binary package has a disabled use flag present but the ebuild doesn't have that use flag present at all, it should ignore the mismatch?
Comment 3 Michał Górny archtester Gentoo Infrastructure gentoo-dev Security 2023-05-17 02:59:31 UTC
There's no rule saying whether flag disappearing means "disabled", "enabled unconditionally" or "something else".  What really needs to happen is your binpkg builder detecting USE change and rebuilding the package.
Comment 4 Jernej Jakob 2023-05-18 11:57:10 UTC
(In reply to Michał Górny from comment #3)
> There's no rule saying whether flag disappearing means "disabled", "enabled
> unconditionally" or "something else".  What really needs to happen is your
> binpkg builder detecting USE change and rebuilding the package.

I was updating my binpkg host with what I thought was correct, 'emerge -auDU @world' but now that I know that, I see I should be using -N instead of -U. So it was my error and this is indeed INVALID.
Maybe a warning about not using -U should be added to the wiki though. I think on pages https://wiki.gentoo.org/wiki/Upgrading_Gentoo#Updating_packages add: "If you are building binary packages, you most likely should not use --changed-use. Also see the [Binary package guide#Maintaining_binary_packages]" and on https://wiki.gentoo.org/wiki/Binary_package_guide#Maintaining_binary_packages add subsection "Updating the binary package host" with the instruction to use the "--newuse" option (-auDN) when updating @world. What do you think? I can edit the wiki if you agree.
Comment 5 Michał Górny archtester Gentoo Infrastructure gentoo-dev Security 2023-05-18 14:05:20 UTC
Yes, please update the Wiki.  I think it makes sense.