#696254 qa.debian.org: PTS has outdated current policy version

Package:
qa.debian.org
Source:
qa.debian.org
Submitter:
Dominic Hargreaves
Date:
2021-02-19 13:27:25 UTC
Severity:
normal
Tags:
#696254#5
Date:
2012-12-18 17:31:14 UTC
From:
To:
The PTS says:

"The package should be updated to follow the last version of Debian
Policy (Standards-Version 3.9.3 instead of 3.9.4)." but this is wrong.

Index: www/xsl/pts.xsl
===================================================================
--- www/xsl/pts.xsl (revision 2895)
+++ www/xsl/pts.xsl (working copy)
@@ -32,7 +32,7 @@

 <!-- Those variables controls the todo and problem item concerning
      standards-version not being up to date -->
-<xsl:variable name="lastsv" select="'3.9.3'"/>
+<xsl:variable name="lastsv" select="'3.9.4'"/>
 <xsl:variable name="lastmajorsv" select="'3.'"/>

 <!-- lintian summary -->

Cheers,
Dominic.

#696254#10
Date:
2012-12-19 05:08:43 UTC
From:
To:
I wonder if we should instead define two versions:

latest: 3.9.4
unstable: 3.9.3

Since policy 3.9.4 is only supposed to take effect after the freeze
period, the PTS should not warn about upgrading to it. It is probably
wrong to warn about using standards-version 3.9.4 though. Could you
implement that?

#696254#15
Date:
2012-12-19 06:37:12 UTC
From:
To:
I suggest to implement the following instead :

The package's standards version should match the version of the package
"debian-policy" in the same distribution as the package.  This should be
checked in distributions testing, unstable and experimental.  If there is no
debian-policy in the distribution, then look at debian-policy in unstable.

- package in testing should match debian-policy in testing
- package in unstable should match debian-policy in unstable
- package in experimental should match debian-policy in unstable

The first three version parts are sufficient, but it's not an error if all four
parts are used.

- debian-policy version 3.9.3.1 --> standards version 3.9.3 is sufficient
- debian-policy version 3.9.4.0 --> standards version 3.9.4 is sufficient

The message should be produced only if the package's standards version is lower
than the (first three parts of the) version of debian-policy, not when it's
higher.  So I would not complain about a package in testing having
standards-version 3.9.4 at this time.

This is the existing message template :

  The package should be updated to follow the last version of Debian Policy
  (Standards-Version <first three version parts of policy version> instead of
  <package's standards version>).

I suggest to change that to this template to allow multiple messages per source
package :

  Version <package version> of this package has "Standards-Version: <package's
  standards version>" and should be updated to follow version <first three
  version parts of policy version> of Debian Policy.

When a package has the same version in multiple distributions, then two
versions of debian-policy could come into play, and then I suggest to look at
the lowest version of debian-policy only.  For example for gwhere
0.2.3.dfsg.1-3 currently in testing and unstable :

  Version 0.2.3.dfsg.1-3 of this package has "Standards-Version: 3.8.1" and
  should be updated to follow version 3.9.3 of Debian Policy.

Regards,

Bart Martens

#696254#20
Date:
2012-12-19 14:59:14 UTC
From:
To:
I understand your proposal, but it feels like overkill to me. I'm not
convinced there's really anything special about 3.9.4 than any other
version of policy w.r.t the freeze. Russ, in his 3.9.4 announcement,
reminded us not to gratuitously update the S-V of a package during
the freeze (and he did mention 3.9.4 there) but wouldn't the same apply
to a package that currently had 3.9.2? A gratuitous bump to 3.9.3
wouldn't make any sense for a package targetted at wheezy.

How about a simpler modification to the current template:

  The package should be updated to follow the last version of Debian Policy
  (Standards-Version <first three version parts of policy version> instead of
  <package's standards version>). We recommend that such changes do not
  take place during a freeze for packages targetted at the testing
  distribution.

#696254#25
Date:
2012-12-19 20:00:37 UTC
From:
To:
It's some work once, but afterwards PTS automatically follows any future
debian-policy version changes, so I think it's worth it.

I don't see anything special there either.

I think he meant that doing uploads and requesting unblocks just to bump the
standards-version is not so useful.

I agree with that.

I agree about "a gratuitous bump".  About "a bump" however, I think that fixing
RC bugs combined with bumping the standards-version without further changes is
not wrong.  It's not wrong that PTS suggests to bump the standards-version,
regardless of the freeze.

I prefer short messages on the PTS.  If we want the PTS to stop suggesting to
bump the standards-version during the freeze, then the messages should be
hidden during the freeze.  Maybe we can do that automatically.  Do we have some
indicator on quantz that Debian is in freeze ?

Regards,

Bart Martens

#696254#30
Date:
2012-12-19 23:05:02 UTC
From:
To:
Right, that part of it makes sense, but I don't think there's any need
to track more than debian-policy in unstable.

I'm not volunteering to implement that, though :)

I don't think the current advice is different to other PTS notices such
as new upstream versions and patch incorporation; they would often
not be appropriate during a freeze. A global reminder of the freeze
and not blindly following the advice of the PTS wouldn't do any harm,
though.

#696254#35
Date:
2012-12-19 23:17:49 UTC
From:
To:
Bart Martens <bartm@debian.org> writes:

The release team have, in the past, asked people not to make changes like
bumping the standards version in changes that need to be reviewed, since
it adds noise to the diff and makes their job harder.  So it's more than
just not uploading solely for standards version changes, but also that one
shouldn't change it in uploads targetted for stable.

I'm happy to go with this policy if the release team is okay with it.  My
understanding is that they'd rather people not do this, which is where my
hesitation comes from.

#696254#40
Date:
2012-12-20 06:14:10 UTC
From:
To:
I think now that you're right about tracking only debian-policy in unstable.

I might volunteer to implement something for this bug when I find the time, but
let's first agree on the desired solution.

You're right that PTS currently produces more messages that should be ignored
during freeze.

That wouldn't do harm, but then it shouldn't be limited to standards-version
only.  Maybe a separate message on the PTS like "Debian is in freeze, so any
uploads for this package most conform to the freeze policy" linking to
http://release.debian.org/wheezy/freeze_policy.html would be a good addition.
We can discuss such global reminder separately from this bug about
standards-version.

Maybe unwanted standards-version bumps have not been much of an issue so far
since I don't see standards-version mentioned in the freeze policy.

Not bumping standards-version at all during freeze corresponds to what I do
with uploads during the freeze, since I don't remember ever having bumped the
standards-version during freeze.

OK, with the feedback I've read so far, I think now that only debian-policy in
unstable should be tracked, doing that automatically would be nice, and it's
not wrong to allow PTS to continue to show messages that should be ignored
during freeze.  If we would want different or less messages on PTS during
freeze, then this goes wider than just the messages about standards-version.
Is this a good summary of what we discussed so far ?

Regards,

Bart Martens

#696254#45
Date:
2017-06-29 04:12:46 UTC
From:
To:
There is a bug in the code for checking a major standards version
increment.  For example, the page for Unifont
(https://packages.qa.debian.org/u/unifont.html) contains this item in
the "Problem" section:

"The package is severly out of date with respect to the Debian Policy.
Latest version is 3.9.8 and your package only follows 4.0.0..."

Note also the misspelling of "severely" as "severly".  The new tracker
spells this correctly.


Paul Hardy

#696254#50
Date:
2021-02-19 13:24:12 UTC
From:
To:
Hallo,

Ich habe dir eine Mail geschickt, aber keine Antwort von dir, warum?

Eddie