nicolas all the information is in the metacello mailing-list and we already posted many mails on that topics in this mailing list. I'm sorry but I do not have the time to repeat again what we said. Now since I'm in a really good mood in a nutshell one repo per version Pharo1.0, Pharo11 ... MetacelloRepository containing configurationOfXXX frozen = all the dependent packages are copied locally + configuration ready to load using load. a project contain its single/or whatever configurationOfMyProject people publish the config to their repository and if they want they push from their repository to the version repository a version that they freeze (may be using tags) For Pharo core or pharo packages that we maintain we want to maintain it as a project. with its own repository and packages. Then we have a server that load configuration and barks if something wrong happen. And may be we kick out the configurationofXXX from MetacelloRepositories
Steph, I understand how boring it may seem: - you already discussed the subject over and over - you already took decisions and now we come later with new proposals, and ask you to reconsider the work again...
more than that. :) It looks like if we did not talk or think about what we are doing.
As for the Preferences/Settings, Pharo choices are probably very weel thought.
However, every solution will come with trade offs, and it would be good to have a rationale justifying the choices you made, and perhaps more importantly justifying why you did abandon some solutions, because some questions might be repeated in Pharo 1.2, 1.3, 2.0 ... and find different answers in different context. Note that Squeak evolutions might have influence on Pharo too, and it's part of the context...
I strongly encourage you to establish your rationale on Pharo grounds - independently of Andreas proposals, put it on the Pharo wiki and let us know the URL.
You also know that Squeak goals are not exactly those of Pharo w.r.t. backward compatibility, so Squeak can't just blindly replicate Pharo solutions without a rationale.
The merit of Andreas proposal is to try and establish such a rationale. Without it, we're bound to repeat discussions again and again,
So to fix the problem consider that as our proposal. Stef