{upgrade} does not expand to 1 even though at least one slot can be upgraded

[Flowdalic]

Shouldn't the output of the following be 'true'

# FORMAT='{upgrade}true{else}false{}' eix --exact --pure-packages sys-kernel/gentoo-kernel
false

given that eix sys-kernel/gentoo-kernel shows

 # eix sys-kernel/gentoo-kernel
[?] sys-kernel/gentoo-kernel
     Available versions:  
     (5.4.242) 5.4.242-r2^tu
     (5.4.243) ~5.4.243^tu
     (5.4.244) 5.4.244^tu
     (5.10.179) 5.10.179-r2^tu
     (5.10.180) ~5.10.180^tu
     (5.10.181) 5.10.181^tu
     (5.15.110) 5.15.110-r2^tu
     (5.15.111) ~5.15.111^tu
     (5.15.112) ~5.15.112^tu
     (5.15.113) ~5.15.113^tu
     (5.15.114) 5.15.114^tu
     (6.1.27) 6.1.27-r2^tu{gpkg}
     (6.1.28) ~6.1.28^tu
     (6.1.29) ~6.1.29^tu
     (6.1.30) ~6.1.30^tu
     (6.1.31) 6.1.31^tu
     (6.2.16) ~6.2.16^tu
     (6.3.4) ~6.3.4-r1^tu
     (6.3.5) ~6.3.5^tu
       {debug hardened +initramfs savedconfig test}
     Installed versions:  6.1.21(6.1.21)^t(10:25:24 2023-04-04)(initramfs -debug -hardened -savedconfig -test) 6.1.22(6.1.22)^t(08:32:58 2023-04-04)(initramfs -debug -hardened -savedconfig -test) 6.1.27-r2(6.1.27)^t{gpkg}(21:39:40 2023-05-30)(initramfs -debug -hardened -savedconfig -test) 6.2.13(6.2.13)^t(10:56:21 2023-04-28)(initramfs -debug -hardened -savedconfig -test)
     Homepage:            https://wiki.gentoo.org/wiki/Project:Distribution_Kernel https://www.kernel.org/
     Description:         Linux kernel built with Gentoo patches

where 6.1.31 is stable and installable? Also note that [?] besides the package name. I would have expected [?U].

[vaeth]

Do you have either set UPGRADE_TO_HIGHEST_SLOT=false or do you have sys-kernel/gentoo-kernel in /etc/portage/package.slot_upgrade_forbid?

[vaeth]

There was a typo on the eix manpage UPDATE_TO_HIGHEST_SLOT (false) instead of UPGRADE_TO_HIGHEST_SLOT (correct).

[Flowdalic]

Do you have either set UPGRADE_TO_HIGHEST_SLOT=false

No, also explicitly setting UPGRADE_TO_HIGHEST_SLOT=true does not change this behavior

$ UPGRADE_TO_HIGHEST_SLOT=true FORMAT='{upgrade}true{else}false{}' eix --exact --pure-packages sys-kernel/gentoo-kernel
false

or do you have sys-kernel/gentoo-kernel in /etc/portage/package.slot_upgrade_forbid

I didn't even know that such a file exists. It appears to be eix specific and does not affect portage (at least I could find the string "slot_upgrade_forbid" in portage's source), right?

[vaeth]

I tried to reproduce the problem by removing my package.accept_keywords for gentoo-sources and lying about the installed version in /var/db/pkg. However, I get [U?] as expected (I did not try the manual {upgrade} check, because this is redundant as being implicitly called when displaying the [U?]).

In the moment I have no idea what might cause the problem for you.

No, also explicitly setting UPGRADE_TO_HIGHEST_SLOT=true does not change this behavior

Please verify that eix --print UPGRADE_TO_HIGHEST_SLOT really prints true. (This is just a desparate attempt, as I really cannot imagine any other cause). Also, please try to regenerate the database with eix-update (perhaps some subtle corruption has happened.)

It appears to be eix specific and does not affect portage

This conjecture is correct. I introduced it as a way to customize UPGRADE_TO_HIGHEST_SLOT per package. Perhaps not the greatest mechanism, but that's how it is now.

[vaeth]

Another possibility (though even more unlikely): Please verify that eix --print SLOT_UPGRADE_FORBID indeed prints the expected path /etc/portage/package.slot_upgrade_forbid.

[Flowdalic]

I just remember what may be the missing piece. I have a user set file /etc/portage/sets/kernel-preserved:

=sys-kernel/gentoo-kernel-6.1.22
=sys-kernel/gentoo-kernel-6.2.13
=sys-kernel/gentoo-kernel-6.1.27-r2
=sys-kernel/gentoo-kernel-6.1.21

maybe this is confusing eix?

[vaeth]

No. eix does not do any dependency resolution. These files are only used to check whether a package is in world{,_sets}. I tried nevertheless, and it had no effect.