[Please don't Cc: me on list mail, as per the Mail-Followup-To: header I set. Thanks.]
On Thu, 07 Dec 2000 at 08:06:17 -0800, Peter Jay Salzman wrote: > On Thu 07 Dec 00, 8:45 AM, Colin Watson said... > > Peter Jay Salzman <[EMAIL PROTECTED]> wrote: > > >i just removed netscape from my system. > > > > > >an error flashed by during the remove, and quickly was replaced by the > > >main dselect menu. i saw enough to know that SOME directory couldn't > > >be deleted, but i don't know which one or why. > > > > Those errors are often spurious, especially if you're purging and > > they're directories containing conffiles that haven't been removed yet. > > not good enough. i _want_ to see them. Didn't say you didn't; just trying to be helpful. > > >at the very least dselect should prompt you to "enter a key" when an > > >error occurs... > > > > Normally in my experience it does, but at any rate you'll notice that > > when you quit dselect the screen with errors on it is restored (unless > > there's something screwy with your terminal, that is). > > not these errors. > > i appreciate the fact that some errors are more serious than others. that > doesn't give dselect the right to enforce its idea of "importancy" on me. > > i maintain that dselect should be showing me these errors. and barring > that, i maintain that it should give me the option of seeing them, which it > currently doesn't. But on my system it does; when I quit dselect the screen containing dpkg output is restored. This is not dselect trying to 'enforce its idea of "importancy"' on anyone, or any such FUD, since the expected behaviour is that when endwin() is called then the previous screen is restored. Granted, on some terminals it doesn't, but that's their fault. What terminal are you using? Does Shift-PgUp happen to scroll back to the error output? Yes, dselect probably ought to have an extra pause for keyboard input in it, but I imagine it would be trivial to fix if you were to tell the maintainers about it rather than debian-user. It's not exactly an irresolvable design flaw. -- Colin Watson [EMAIL PROTECTED]
