Question on why package was rebuilt

[Loren M. Lang]

Hello,

I recently had a package sponsors and entered into unstable called tiv.
It can be seen here:

https://packages.debian.org/sid/tiv

Everything went OK, but I see that the amd64 arch package appears to
have been re-built for some reason. It's version is showing up with a
+b1. I am curious if there is some long to indicate what the issue might
have been that led to a rebuild. Could there have been a compilation
issue or other things I should be concerned about or is it likely
something harmless? Is there a log for this case?

--
Loren M. Lang
lor...@north-winds.org
http://www.north-winds.org/


Public Key: http://www.north-winds.org/lorenl_pubkey.asc
Fingerprint: 7896 E099 9FC7 9F6C E0ED E103 222D F356 A57A 98FA

[Mathias Gibbens]

On Thu, 2024-02-15 at 16:20 -0800, Loren M. Lang wrote:
> Hello,
>
> I recently had a package sponsors and entered into unstable called tiv.
> It can be seen here:
>
> https://packages.debian.org/sid/tiv
>
> Everything went OK, but I see that the amd64 arch package appears to
> have been re-built for some reason. It's version is showing up with a
> +b1. I am curious if there is some long to indicate what the issue might
> have been that led to a rebuild. Could there have been a compilation
> issue or other things I should be concerned about or is it likely
> something harmless? Is there a log for this case?

There's no cause for concern -- it's a normal part of a new package
entering the archive.

When a package is uploaded to NEW, you have to upload both the source
and binary package(s) for review. After the package is accepted, the
buildds auto-build for any other architectures that don't already have
a binary package. Migration policy requires all packages to be built on
official buildds from their source package[1]. Since the amd64 binary
package already existed from the upload to NEW, it wouldn't be auto-
built and would block migration of your package to testing.

The "+b1" indicates a binBMU was performed[2,3]. If you look at the
buildd logs (https://buildd.debian.org/status/package.php?p=tiv),
you'll see the relevant changelog entry for the amd64 package: "Rebuild
on buildd".

Mathias

[1] -- https://www.debian.org/doc/manuals/developers-reference/pkgs.en.html#source-and-binary-uploads
[2] -- https://www.debian.org/doc/manuals/developers-reference/pkgs.en.html#source-nmus-vs-binary-only-nmus-binnmus
[3] -- https://www.debian.org/doc/manuals/developers-reference/pkgs.en.html#recompilation-or-binary-only-nmu

[Sebastian Ramacher]

Hi

On 2024-02-16 21:32:49 +0100, Jérémy Lal wrote:
> When a package is uploaded to NEW, you have to upload both the source
> > and binary package(s) for review. After the package is accepted, the
> > buildds auto-build for any other architectures that don't already have
> > a binary package. Migration policy requires all packages to be built on
> > official buildds from their source package[1]. Since the amd64 binary
> > package already existed from the upload to NEW, it wouldn't be auto-
> > built and would block migration of your package to testing.
> >
>

> This isn't what happened, I suppose, since we all debian maintainers need
> to do source-only uploads after a package has been accepted through the NEW
> process.
> Unless I'm mistaken, that source-only upload cannot be replaced by a
> binNMU, can it ?
> What happened is more likely to be a standard rebuild against a new version
> of a dependent library.

A binNMU is enough if the source package only builds architecture
dependent packages. If Architecture: all packages are involved, a source
only upload is required.

Cheers
--
Sebastian Ramacher

[Jérémy Lal]

Le ven. 16 févr. 2024 à 04:03, Mathias Gibbens <gib...@debian.org> a écrit :

On Thu, 2024-02-15 at 16:20 -0800, Loren M. Lang wrote:
> Hello,
>
> I recently had a package sponsors and entered into unstable called tiv.
> It can be seen here:
>
> https://packages.debian.org/sid/tiv
>
> Everything went OK, but I see that the amd64 arch package appears to
> have been re-built for some reason. It's version is showing up with a
> +b1. I am curious if there is some long to indicate what the issue might
> have been that led to a rebuild. Could there have been a compilation
> issue or other things I should be concerned about or is it likely
> something harmless? Is there a log for this case?

  There's no cause for concern -- it's a normal part of a new package
entering the archive.

Indeed...

  When a package is uploaded to NEW, you have to upload both the source
and binary package(s) for review. After the package is accepted, the
buildds auto-build for any other architectures that don't already have
a binary package. Migration policy requires all packages to be built on
official buildds from their source package[1]. Since the amd64 binary
package already existed from the upload to NEW, it wouldn't be auto-
built and would block migration of your package to testing.

This isn't what happened, I suppose, since we all debian maintainers need

to do source-only uploads after a package has been accepted through the NEW process.

Unless I'm mistaken, that source-only upload cannot be replaced by a binNMU, can it ?

What happened is more likely to be a standard rebuild against a new version of a dependent library. 

  The "+b1" indicates a binBMU was performed[2,3]. If you look at the
buildd logs (https://buildd.debian.org/status/package.php?p=tiv),
you'll see the relevant changelog entry for the amd64 package: "Rebuild
on buildd".

A binNMU, but right.