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

Re: testing is broken



On Thu, Apr 05, 2001 at 11:54:32PM -0700, Thomas Bushnell, BSG wrote:
[snip]

> 
> But it should not be suggested to the new user, and the few things
> that apt still doesn't do now, should be priorities.  (The real biggie
> is dealing with suggestions and enhances.)
> 
> One interface would be:
> 
> $ apt-get install foo
> Package foo suggests also installing package bar; install it too
> (y/N)?
> 
> Pretty simple and functional.
> 
When going to implement that I would say that you will run into the
same problems with dselect, if you answere Y to some of the suggest
questions (package A suggest B, B conflicts C, and you have C selected
to be installed). I'd say if that happens then you would have to answere
another question detailing the upcoming problem like:
Package foo suggest installing package bar; bar conflicts foobar (which
is already or going to be installed) Do yo really want to install bar (y/N)?

Several other ways might be there to avoid this like:
Never ask when conflicts arise, Just ask one question similar to the one
above (this sollution I would prefer) 

Yeah another thing that might arise is that people simply don't like that
additional suggest thingy and want to turn it off or they want to turn it
partionally off e.g. Conflicts should never be asked after, if the packages
that are involved are already installed or have been specified by the user.

So that user settings have to be stored somewhere as the answeres should be
stored on a per package basis.

We all can see now that it is not trivial to solve that problem and
therefore it was not addressed in apt-get til now.

Personal I don't care about that suggest thing, cause I don't mind to test
my system for functionality and install missing packages by hand. And if
this suggest thing is going to be a standard feature of apt-get that I
cannot turn off, I start to use dpkg directly. Beeing asked silly suggest
questions all time will grow annoying I think and that's not worth it.

-- 
kind regards, 
Michael Moerz 

pub  1024D/B651C436 2000-09-17 Michael Moerz <e9625136@stud3.tuwien.ac.at>
     Key fingerprint = 55DB 2F1A BF45 DBAB F542  4128 2173 8753 B651 C436
http://idc19.itm.tuwien.ac.at/~mike/private/mike.public.gpg.key

Attachment: pgpR5VJZasGrW.pgp
Description: PGP signature


Reply to: