Documentation updates in-cycle?

[Matthew East]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi,

* Duncan Lithgow:
> First I should check my facts: is it correct that corrections to
> released documentation do not get released and pushed/ made available to
> users?
> 
> If that's the case I don't understand why.
> 
> I understand that the Debian policy (which ubuntu has adopted) is that
> only security updates get pushed down to users. And I think that's fine
> - I assume that's to avoid introducing new bugs and all the hassle that
> comes with that.
> 
> But I think documentation is a totally different thing, and by this I
> mean strictly the text and images and not the underlying code (excluding
> any markup).
> 
> Surely it is as good as impossible to introduce bugs by simply making
> changes to documentation as defined above. (Further documentation
> mistakes perhaps, but not real bugs)
> 
> I understand that there are other issues about how corrected docs would
> be distributed - but let's leave those issues for now and restrict
> ourselves to the question:
> 
> "Are there enough advantages to correcting docs that we should consider
> looking at a way to do it?"

You're right that currently documentation is subject to the same
policies as all of Ubuntu - https://wiki.ubuntu.com/StableReleaseUpdates

If we decide that we'd like an exception to be carved out of the
StableReleaseUpdates policy, we'll need to approach the Technical Board
with some fairly cogent arguments.

I think there are three considerations to balance:

1. the risk of introducing new bugs - it's possible to introduce real
bugs, although I doubt they would be that serious - for example, you
might write some invalid markup that results in scrollkeeper errors that
end up in people's monthly cronjob error logs. I personally can't think
of anything more serious than that (and even that would be quite
difficult to do) but I'm not an expert with Ubuntu and I'm told that
bugs can come from unexpected places;

2. the resources taken up by doing the update - when doing an update
it's necessary to: (a) prepare a fix both for the unstable version of
Ubuntu and each stable version to which the bug applies; (b) make sure
the translators know about the change and can translate it; (c) import
the relevant translations; (d) prepare an updated package and submit it
to the relevant developer team for checking. The documentation team is
very small and 100% volunteer based and we don't have much time.

3. the seriousness of the bug.

My personal view is that *probably* when balancing all these things up,
the bug would have to be high-impact before it justifies undertaking the
process to fix it, which takes us back to the StableReleaseUpdates policy.

What do other people think? I'm cc:ing Tollef for his views.

Matt
- --
http://www.mdke.org
gnupg pub 1024D/0E6B06FF
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFGP178tSaF0w5rBv8RAgVkAJ0XbZxTbpYyZIjCsYp8lBQAHFiK6QCeNuH5
sxUPUmN3BDwGVi5YEvXlz2w=
=qiSn
-----END PGP SIGNATURE-----