[php-maint] 5.4.0~rc6-2 (in experimental) testers needed (packaging, not php5 yet)

[Ondřej Surý]

On Thu, Feb 2, 2012 at 10:26, sean finney <seanius at debian.org> wrote:
> On Thu, Feb 02, 2012 at 08:40:38AM +0100, Ondřej Surý wrote:
>> We're not, but this approach is much simpler and doesn't break the existing
>> non-pkg-php packages (they can still use conf.d).
>
> right, and there's some argument for that.  note that i'm not actually
> objecting to going that route, just that way back years ago when we
> actually did spend braincells thinking about this we settled on a
> different way and want to make sure that if we decide to drop the
> extra ideas that it's intentional.

Intentional in this case means 'something we have manpower and time to do'.

> it is still possible, after all, to configure it manually per-sapi
> by breaking the per-sapi conf.d symlink.
>
> we could also, for example (i have not really thought hard on this), add
> the per-sapi options to the existing script, but not fiddle with the conf.d
> directories.  i.e. so that it *would* work in the case that the sapi conf.d
> were manually made distinct by the admin, and otherwise spit out a snarky
> diagnostic pointing at README.Debian or a manpage or something.

My grand schema included templates and nice interface to check which
modules you want to have enabled (and remember that in debconf database).
Some years later... I am glad I forced myself to do at least what I did now.

>> Yes, it would, you're right here. My excuse is that I was in packaging
>> rage yesterday :)
>>
>> But frankly when was the last time you or Raphael has touched the repo
>> last time? I have given up the hope :(.
>
> haven't touched the repo in a while, but i'm trying to at least remain
> responsive for the significant points of discussion

I know and I thank you for doing that.

>> > Also, debian/copyright should probably be updated, since canonical has
>> > claimed copyright on the script :)
>>
>> Yeah, I need to speak with Clint about that yet.
>
> Like the other matter, it wasn't raised as an objection per se (if they
> want to copyright+GPLv3 it, i personally find it a bit silly but not
> a problem).

I agree on the silly part, so I want to speak with Clint...

> Just, we need to be good citizens and reflect that in
> debian/copyright like we do for the other significant parts of the
> packaging.

Agree.

>> Well, I think that Lior's hint about using ucf was good one. It just
>> seemed to me
>> that it would be too much, when I first though about that. But I'll
>> throw some magic
>> ucf dust over php5-module.*
>
> my biggest complaint with ucf is that it's dog slow, but otherwise it's
> a step up from dpkg's conffile handling imo.  In that case, we wouldn't need
> the versioned Depends on dpkg for the maintscript helper.  But I guess
> we'd still need to pay attention to preserve the pre-existing state
> of the conffiles (modified or removed) in the resulting ucf-registered
> files at the new location.

I think I managed to cover most cases by running ucf and dpkg-maintscript-helper
in right order, but need to carefully check it. I will push the result
afterwards to repo
and probably bundle it with 5.4.0RC7.

O.
-- 
Ondřej Surý <ondrej at sury.org>