Fixes #1884 - hardcode a version number for Debian's testing/unstable suites

[olasd]

Here's the patch as attached to http://theforeman.org/issues/1884

[ohadlevy]

thanks @olasd!, @GregSutcliffe mind having a look?

[skottler]

@olasd I just tested this by running lsb_release -r -s and on unstable, for example, I get unstable as the output.

Since unstable and testing are always in that state (i.e. unstable never directly turns into sid), I think this will cause some confusion once sid final is shipped.

[domcleal]

The -r flag is lsbdistrelease, not lsbdistcodename (used in the patch), but that said, lsbdistrelease appears to return the version number 7.0 on Debian testing currently, which is what we need (without a lookup table). unstable will always be sid, which can always be 99.

How about:

if os_name == "Debian"
  case orel
    when "testing"
      orel = facts[:lsbdistrelease]
    when "unstable"
      orel = "99"
  end
end

[olasd]

On dim., janv. 06 2013, Dominic Cleal wrote:

The -r flag is lsbdistrelease, not lsbdistcodename (used in the
patch), but that said, lsbdistrelease appears to return the version
number 7.0 on Debian testing currently, which is what we need
(without a lookup table). unstable will always be sid, which can
always be 99.

Hmm. It seems that wheezy only got this number a few days ago (in
base-files version 7.1, migrated to testing on 2012-12-26). Before that,
lsbdistrelease was "testing". AFAICT, the version number is always set
at the very end of the release cycle (at least it has been for the past
few cycles).

How about:

if os_name == "Debian"
  case orel
    when "testing"
      orel = facts[:lsbdistrelease]
    when "unstable"
      orel = "99"
  end
end

Hmm, orel is set to lsbdistrelease already, so the testing case is a
noop. In fact, right now my patch is a noop for up-to-date wheezy, as it
only looks for testing or unstable in lsbdistrelease.

I'm not sure what to do next. The bug report is moot for current
testing, but will show up again as soon as wheezy is released (and
someone want to install testing).

Thoughts?

Ever heard of .cshrc?
That's a city in Bosnia. Right?
(Discussion in comp.os.linux.misc on the intuitiveness of commands.)

[domcleal]

On 06/01/13 23:59, Nicolas Dandrimont wrote:

On dim., janv. 06 2013, Dominic Cleal wrote:

The -r flag is lsbdistrelease, not lsbdistcodename (used in the
patch), but that said, lsbdistrelease appears to return the version
number 7.0 on Debian testing currently, which is what we need
(without a lookup table). unstable will always be sid, which can
always be 99.

Hmm. It seems that wheezy only got this number a few days ago (in
base-files version 7.1, migrated to testing on 2012-12-26). Before that,
lsbdistrelease was "testing". AFAICT, the version number is always set
at the very end of the release cycle (at least it has been for the past
few cycles).

Ack, bad time for me to test.

How about:

if os_name == "Debian"
case orel
when "testing"
orel = facts[:lsbdistrelease]
when "unstable"
orel = "99"
end
end

Hmm, orel is set to lsbdistrelease already, so the testing case is a
noop. In fact, right now my patch is a noop for up-to-date wheezy, as it
only looks for testing or unstable in lsbdistrelease.

Oh, you're right - I thought it was just operatingsystemrelease.

Sorry for the noise!

Dominic Cleal
Red Hat Engineering

[GregSutcliffe]

@skottler sid is never released. It's the equivalent of rawhide. Once a package has been in sid for a while it moves to the current testing release.

@olasd looks pretty awesome to me. I agree that we can;t really avoid a lookup table, since the version number only appears at the end of the release cycle. It's not too much work to maintain it, given Debian's release cycle :)

@ohadlevy not tested, but looks fine to me 👍

[skottler]

@GregSutcliffe makes sense. Since Debian releases relatively rarely I don't think this will become a maintenance burden; just would have been nice to avoid maintenance altogether.

Anyhow, looks good style and code-wise to me in it's current state so 👍

[ohadlevy]

merged - thanks @olasd!