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

Bug#826587: apt: misleading message on trying to remove Important:yes packages



On Mon, Jun 06, 2016 at 09:23:30PM +0200, Adam Borowski wrote:
> On Mon, Jun 06, 2016 at 08:48:12PM +0200, Julian Andres Klode wrote:
> > On Mon, Jun 06, 2016 at 07:29:44PM +0200, Adam Borowski wrote:
> > > When trying to remove an Important:yes package, the message claims it's
> > > essential:
> 
> > Well, it sort of is. Not in the "Essential: yes" sense, but in a broader
> > sense.
> 
> My point is, the word "essential" has a specific meaning, so reusing it for
> something only somewhat similar is bound to cause confusion -- at the least,
> make the user think he's dealing with an actually Essential package.

The same applies to important, which is the main reason everyone wants a 
different name :/ (it's a historic artefact that I just re-used, it used
to mean the same as Essential in early APT versions; now it's basically: Do
not remove if already installed).

> 
> There's a difference between needing the highest level of confirmation known
> to apt and something dpkg doesn't even require confirmation for.
> 
> > we have not decided on an official name for that field, the current one is
> > just a very old one.
> 
> In that case, perhaps using it in packages is premature?

Not really, we're (I am) supporting it. It's mostly a matter of getting dpkg to support
it for extra safety that an official field would bring (dpkg would then require
--force-remove-somethinglikeessential to remove it).

Main use case so far was local system configuration meta packages, but init
systems and bootloaders seem like a very good thing as well. Basically anything
that should not be removed normally once installed.

> 
> > So, no idea what to do here.
> 
> Hmm, if you're still debating what Important should do (and even its very
> name), making big changes here might indeed be a waste of effort because of
> the risk of having to do them again.  Thus, what about changing just the
> message for now?  An interface for overriding it can wait until you're happy
> with the specs.

I'm not sure what you mean with overriding it - my problem is to
think of a more useful message...

-- 
Debian Developer - deb.li/jak | jak-linux.org - free software dev

When replying, only quote what is necessary, and write each reply
directly below the part(s) it pertains to (`inline'). Thank you.


Reply to: