Re: Using ports across major version changes

On Dec 17, 2025, at 06:30, bob prohaska <[email protected]> wrote:

> On Tue, Dec 16, 2025 at 05:45:03PM -0800, Mark Millard wrote:
>> bob prohaska <fbsd_at_www.zefox.net> wrote on
>> Date: Tue, 16 Dec 2025 23:47:14 UTC :
>> 
>>> On Tue, Dec 16, 2025 at 08:49:06PM +0100, Dag-Erling Smørgrav wrote:
>>>> bob prohaska <[email protected]> writes:
>>>>> What's the best way to restore normal operation? Something like 
>>>>> poudriere bulk -a
>>>> 
>>>> poudriere bulk $(pkg query -e '%#r == 0' '%o')
>>>> 
>>> 
>>> Something's amiss. A simple copy-paste of the command yields
>>> "Illegal variable name.", maybe I'm using the wrong shell.
>> 
>> And older --and possibly less shell dependent-- notation
>> would be the use of a pair of backquotes:
>> 
>> poudriere bulk `pkg query -e '%#r == 0' '%o'`
>> 
>> (That notation is not so nice for usage that involves
>> wanting nested usage.)
>> 
>>> However, it appears that pkg query -e '%#r == 0' '%o'
>>> generates a list of built packages, so I tried running
>>> pkg query -e '%#r == 0' '%o' > package.list
>>> which worked, followed by
>>> poudriere bulk -j main -f package.list
>>> 
>>> That got off to a good start but didn't end well:
>>> 
>>> [00:00:01] Creating the reference jail... done
>>> [00:00:54] Mounting system devices for main-default
>>> [00:00:54] Mounting ports/packages/distfiles
>>> [00:00:54] Stashing existing package repository
>>> [00:00:58] Mounting packages from: /usr/local/poudriere/data/packages/main-default
>>> /etc/resolv.conf -> /usr/local/poudriere/data/.m/main-default/ref/etc/resolv.conf
>>> [00:00:58] Starting jail main-default
>>> [00:01:10] Logs: /usr/local/poudriere/data/logs/bulk/main-default/2025-12-16_15h32m41s
>>> [00:01:10] Loading MOVED for /usr/local/poudriere/data/.m/main-default/ref/usr/ports
>>> [00:01:25] Ports supports: FLAVORS SUBPACKAGES SELECTED_OPTIONS
>>> [00:01:25] Gathering ports metadata
>>> [00:01:25] Error: MOVED: devel/libpthread-stubs 2023-03-12 No consumers left and never supported pthread stubs in libc on FreeBSD
>>> [00:01:25] Warning: MOVED: devel/pygobject3-common renamed to devel/pygobject-common
>>> [00:01:25] Warning: MOVED: net/openldap24-client renamed to net/openldap25-client
>>> [00:01:25] Error: MOVED: x11-fonts/gentium-basic 2025-12-04 Has expired: Superceeded by Gentium-7.000 https://software.sil.org/gentium/download/

Turns out the above line was because the MOVED line for it
was incorrect. That has been fixed to also list
x11-fonts/gentium as what to use instead:

Wed, 17 Dec 2025

    • git: 24b7568ff29c - main - MOVED: x11-fonts/gentium-basic is superceeded by x11-fonts/gentium Matthew Seaman

>>> [00:01:25] Warning: MOVED: x11-themes/kf5-oxygen-icons5 renamed to x11-themes/oxygen-icons
>>> [00:01:25] Error: Fatal errors encountered gathering initial ports metadata
>>> [00:01:25] Cleaning up
>>> [00:01:25] Unmounting file systems
>>> root@nemesis:/usr/local/poudriere # 
>>> 
>>> The reaction to errors is surprising; it seems like both have enough context to handle.
>>> The first package can be omitted, and the second package can be replaced.
>> 
>>> 
>>> What am I missing?
>> 
>> 
>> Sounds like you can remove devel/libpthread-stubs from
>> package.list and can pkg delete it?
>> 
> 
> Yes to both. 
>> Sounds like you can replace x11-fonts/gentium-basic with
>> x11-fonts/gentium in package.list and can:
>> 
>> pkg delete x11-fonts/gentium-basic
>> 
> Yes to both.
> 
>> Then try the bulk build again based on the updated file.
> 
> Looks like it's working.....Oops, maybe not:
> 00:11:32] [03] [00:02:29] Saved devel/autoconf | autoconf-2.72 wrkdir to: /usr/local/poudriere/data/wrkdirs/main-default/default/autoconf-2.72.tbz
> [00:11:37] [03] [00:02:34] Finished devel/autoconf | autoconf-2.72: Failed: run-depends
> [00:11:49] [03] [00:02:46] Skipping graphics/GraphicsMagick | GraphicsMagick-1.3.43_3,1: Dependent port devel/autoconf | autoconf-2.72 failed
> followed by a torrent of "Skipping.....[various port names]"
> lines.
> 
> I'll let it run until advised to intervene. For now it doesn't look promising. 

Failed: run-depends

The autoconf-2.72 related poudriere build log file will need
to be investigated for evidence about that.

> Thank you for the guidance. I've never seen poudriere complain about
> changes or invite manual intervention, thus my caution about tampering.
> The manual changes don't look obviously related to the apparent problems.


===
Mark Millard
marklmi at yahoo.com