[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Draft spec for new dpkg "triggers" feature (v2, repost)



We're getting there. :)

Ian Jackson <ian@davenant.greenend.org.uk> wrote:

> > I'm not sure what "relevant" means here. E.g., if I do:

[...]

> It will try to process triggers activated during that run.  This
> includes:
>   - triggers activated directly by foo
>   - triggers activated by something else which happens in this run
>      (eg deconfiguration or removal of some other package)
>   - triggers in foo activated by other trigger processing being
>      performed
>
> It generally doesn't include any triggers foo is interested in because
> during most of this operation foo in states similar to unpacked; foo's
> postinst is supposed to do whatever might be needed.  The only case
> where this doesn't apply is where some other package's trigger
> processing, caused by the installation of foo and typically done after
> foo is configured, itself activates one of foo's triggers.  Then foo's
> postinst would be run twice (or perhaps more times).

OK, that's clear. Did you clarify that in the text? Maybe the paragraph
I quoted just above is too much focused on explaining a corner case, but
at least a sentence like your "It will try to process triggers activated
during that [dpkg] run" seems necessary to me.

> apt doesn't do that.  It configures packages in batches.  And indeed
> for other reasons it ought to be taught not to try to tell dpkg which
> packages to configure.

Well, that's another debate I'm not really qualified to participate
in...

>> > [ state diagram ]
>> 
> The difficulty with that is that triggers-awaited is a bit of an
> anomaly.  It doesn't fit in just one box on the diagram - it overlays
> onto triggers-pending or installed.  But I think it ouught to be
> mentioned there so I have updated the diagram.

OK.

>> Hmm, during my last upgrades, I've seen messages like:
>>   Deferring scrollkepper update...
>> (quoting from memory; maybe the phrasing is different), so I'm wondering
>> whether it is still true that "currently this occurs at every relevant
>> package installation, upgrade or removal".
>
> Perhaps it defers it if scrollkeeper itself isn't configured or some
> other similar situation exists ?

Frankly, I have no idea. To be sure, one should have a close look at the
relevant maintainer scripts or more simply ask the scrollkeeper
maintainers...

> If you don't give the --by-package option and you're not a child of
> dpkg (which will have set an environment variable for dpkg-trigger's
> benefit) then dpkg-trigger will fail.

OK. Thanks!

-- 
Florent



Reply to: