Hello.

As the subject says, many postgresql-related packages FTBFS in this way:
(I'm taking postgresql-unit as an example)

make[1]: Leaving directory '/<<PKGBUILDDIR>>'
dh_clean
pg_buildext checkcontrol
--- debian/control 2024-09-15 11:58:15.000000000 +0000
+++ debian/control.VFAeRh 2024-09-19 05:00:18.737341291 +0000
@@ -14,7 +14,7 @@
Vcs-Browser: https://github.com/df7cb/postgresql-unit
Homepage: https://github.com/df7cb/postgresql-unit

-Package: postgresql-17-unit
+Package: postgresql-16-unit
Architecture: any
Depends: ${misc:Depends}, ${shlibs:Depends}, ${postgresql:Depends}
Description: SI Units for PostgreSQL
Error: debian/control needs updating from debian/control.in. Run 'pg_buildext updatecontrol'.
If you are seeing this message in a buildd log, a sourceful upload is required. make: *** [debian/rules:10: clean] Error 1 shuffle=2080555641 dpkg-buildpackage: error: debian/rules clean subprocess returned exit status 2


Is this really normal/expected, or is there a bug somewhere? (like, for example,
a missing build-dependency which propagated to testing prematurely).

I'm concerned about this because even if this is the kind of bug which
end up being fixed ove

On Wed, Sep 25, 2024 at 03:09:35PM +0200, Santiago Vila wrote:

Hello.

As the subject says, many postgresql-related packages FTBFS in this way:
(I'm taking postgresql-unit as an example)

make[1]: Leaving directory '/<<PKGBUILDDIR>>'
dh_clean
pg_buildext checkcontrol
--- debian/control 2024-09-15 11:58:15.000000000 +0000
+++ debian/control.VFAeRh 2024-09-19 05:00:18.737341291 +0000
@@ -14,7 +14,7 @@
Vcs-Browser: https://github.com/df7cb/postgresql-unit
Homepage: https://github.com/df7cb/postgresql-unit
-Package: postgresql-17-unit
+Package: postgresql-16-unit
Architecture: any
Depends: ${misc:Depends}, ${shlibs:Depends}, ${postgresql:Depends}
Description: SI Units for PostgreSQL
Error: debian/control needs updating from debian/control.in. Run 'pg_buildext updatecontrol'.
If you are seeing this message in a buildd log, a sourceful upload is required.
make: *** [debian/rules:10: clean] Error 1 shuffle=2080555641 dpkg-buildpackage: error: debian/rules clean subprocess returned exit status 2

Is this really normal/expected, or is there a bug somewhere? (like, for example,
a missing build-dependency which propagated to testing prematurely).

This is expected for PostgreSQL transitions.

I'm concerned about this because even if this is the kind of bug which
end up being fixed over time, if we were to release trixie as stable today, we would ship many FTBFS bugs like those, so this is apparently something that we should avoid to happen.

This is tracked as an ongoing transition, which is already mostly
finished in unstable:
https://release.debian.org/transitions/html/postgresql-17.html

E.g. for postgresql-unit in your example the renamed package for 17 has
already migrated - but postgresql-common which changes the default not yet.

Thanks.

cu
Adrian

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Re: Santiago Vila

Is this really normal/expected, or is there a bug somewhere? (like, for example,
a missing build-dependency which propagated to testing prematurely).

As Adrian already said, everything is normal and under control. As
opposed to "normal" library transitions where everything needs to
transition at once, PostgreSQL transitions allow packages to migrate
one after another, with almost zero release team intervention. (This
time it's a bit more since we are dropping 32-bit architectures and
some arch:all packages need hinting.)

I'm concerned about this because even if this is the kind of bug which
end up being fixed over time, if we were to release trixie as stable today, we would ship many FTBFS bugs like those, so this is apparently something that we should avoid to happen.

Well, we are not going to release trixie in the next two weeks.
PostgreSQL 17.0 will be out tomorrow and I expect postgresql-common to
migrate very soon; I uploaded fixes for two php and django packages
today. This will flip the set of supported versions.

Christoph

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Thanks for the hints. It seems postgresql-common will finally propagate
to testing this night:

https://tracker.debian.org/pkg/postgresql-common

[ For the curious: I still do archive rebuilds mostly in testing, but I always double-check in unstable before reporting, so these anomalies are some kind
of "false-positives" in my setup. Will look at current transitions first
if a similar case arises ].

Thanks.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)