On Wed, 2002-07-31 at 14:52, Anthony Towns wrote: > On Wed, Jul 31, 2002 at 03:01:30PM +0200, Raphael Hertzog wrote: > > [Christian] wrote 3 conversion scripts for gnome-terminal, gnome-session and > > gnome-panel. > > > > The "problem" above should've specified that. AIUI, everyone believes > > > Gnome2 will be ready for sarge and sid at the point when an upgrade from > > > 1.4 to 2.0 preserves most of the configs of all the users on the machine. > > There's no evidence that this may automatically happen one day. > > Of course it won't happen "automatically". It'll happen precisely when people > get out there and write the upgrade scripts. > > Now, personally, I'd've thought the Gnome2 hackers should've written > those scripts before they released Gnome2. But they didn't, and if they > don't care about our users, well, that's our job anyway. There're two > possibilities: we either say the same thing that Gnome upstream did, ie > "Your configuration settings aren't considered important. If you liked > 'em, recreate 'em. It's all too much effort for us." and put Gnome2 into > unstable as is, or we say "Gnome upstream are completely incompetent and > should never have released Gnome2 yet. Unlike them, we're not incompetent, > and we're going to spend some time doing the work those lazy sonsabitkas > should've done, and make sure your upgrades are as smooth as they can be." Stop saying BS like that. The upstream has never *ever* said that "configuration settings aren't important". Jeff has telling people that ever since G2D was effectively released. We were hoping that distributors of Gnome2 (mainly Ximian, Redhat and Mandrake actually) would help us out with that. If we didn't release G2D and push for it in in the distros, the resources wouldn't have been allocated for this work to be done, and G2D would have taken even longer to arrive. And I certainly don't appreciate the people that put so much work into making Gnome2 or myself being called "lazy sonsabitkas". Say that on IRC while I'm looking away, but recording it in the BTS is the icing. > If we want to do the former, then we replace Gnome1.4 in unstable, > and let Gnome2 roll into testing in the usual fashion. > > If the latter's considered better -- which it certainly seems to be to me, > although I gave up on Gnome configuration ages ago -- then people need to > setup somewhere that they can *write* such scripts, and work out a way of > delivering them. This can happen after everyone with unstable installed > has lost all their settings, but that'd be something of a waste, at least. Actually, we would prefer changes to be done to the C code in the applications themselves, so that each and every application knows where to stand wrt migration. For mass transition, scripts are fine though. Maybe they could be hooked up in the gdm/kdm login scripts. It would be good to hear Ryan on that, as he maintains gdm. > Updating people's settings doesn't have to be too clever or automatic, > if that's difficult. I don't know enough about the way Gnome1.4 / Gnome2 > to make definite comments, but a fairly good arrangement would be something > like: > > * admin installs gnome2.0 > * user who used to run gnome1.4 logs in, finds spartan desktop, > sees an icon "Reconfigure based on my old Gnome 1.4 settings" > * user double clicks, runs scripts, gets told to log out and log > back in again > * user does so, sees something not completely repulsive Maybe they could be hooked up in the gdm/kdm login scripts. It would be good to hear Ryan on that, as he maintains gdm. Font configuration needs some love as well. The default font configuration, without the Microsoft Truetype fonts is quite horrible and we really can't ask people to install non-free software to make a free software desktop look alright. > Note that supporting partial upgrades makes this difficult: if you just > upgrade gnome-terminal, your desktop won't be very spartan but you'll > still have lost your gnome-terminal settings. Just dumping all of Gnome2 > in unstable on a single day could "optimise" this so that few enough > people are affected that it's not worth caring. > > Anyway, the point of a staging area is to give you somewhere to do > this development until you get to a point where it's ready for public > consumption, nothing more. Breaking unstable is fine, just try to avoid > breaking it in ways that you know can't be fixed for quite a while. We used experimental as the staging area. It would be nicer to use that rather than an outside area, as people can upload binaries for their architectures in a simple, automated way. Cheers -- /Bastien Nocera http://hadess.net
Attachment:
signature.asc
Description: This is a digitally signed message part