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

Re: dselect improvements



On 17-Oct-99, 13:24 (CDT), Milan Zamazal <mz@pdm.pvt.net> wrote: 
> 
>     SG> 1. Only check recommends/suggests when the status of a package
>     SG> changes (i.e. unselected->selected). The continual
>     SG> prompting/overriding is horribly annoying.
> 
> I don't understand this.  When is the check performed except of status
> change requests and quitting the selection?

Suppose I select previously unselected package X, which recommends Y
and suggests Z. I get taken to the "recursive package listing" (RPL),
which I think of as the "conflict resolution screen". So in the RPL, I
find Y selected. I don't want Y, possibly because the maintainer of X
has overstated the importance of Y, possibly because I have specialized
needs, possibly just to be contrary. So I hit 'R' to revert to my
original selection (X yes, Y and Z no), and then 'Q' to get back to the
normal dselect browser. All well and good and reasonable.

So I browse a little more and maybe select a few other packages, maybe
not. I press <return> to exit the browser and go back to the menu to
install my lovely new packages. Whoops! Dselect has noticed that X
recommends Y, and takes me back to the RPL. Once again 'R' and 'Q' do
the trick, and I can go on.

Ok, next day, start dselect, U)pdate, S)elect, browse a bit, <return>
and I'M BACK IN THE DAMNED RPL, because once again dselect *really*
wants me to install Y, because it is recommended by X, even though I've
already had several opportunities to select it and have rejected that
choice EACH AND EVERY TIME.

And so it goes, until I finally get tired of having to futz with it,
and I either install Y, even though I have no use for it, or add it to
equivs.

So, my point was that dselect should only check recommends and suggests
when the status of the package doing the r/s actually changes -- I hit
"+" or "-". I'd tolerate during upgrades of the packages. But this every
time thing has got to stop.


Steve

-- 
Steve Greenland <vmole@swbell.net>
(Please do not CC me on mail sent to this list; I subscribe to and read
every list I post to.)


Reply to: