Thanks for the answer, Keith.
I just realized that your goals and those of the Pharo leaders seem to overlap quite a lot. Small base image, automatic build, build up images with scripts from packages. The main difference being backwards compatibility, right?
Cheers, Bernhard
Am 28.06.2009 um 22:17 schrieb Keith Hodges:
Dear Juan,
Ok. So, what does that set of objectives and agenda say about:
- Backwards compatiblity
We have had Installer, and LevelPlayingField for ages. The purpose of which is to make some compatability possible between different images, and ongoing changes to API's can be back ported to other images thus preserving backwards compatability, and providing forwards compatability where possible.
We have more backwards and forwards compatibility now than we ever had before.
The crux of the difference between squeak and pharo philosophies has been the issue of backwards compatability.
- Removal (or not) of stuff
All tools we are using for 3.11 are not reliant upon a gui. This is explicitly because the stated goal has been for at least 3 years to work towards a kernel image, with which people can build up the custom image that they want.
- Addition (or not) of stuff
The proposal is to have a range of deliverables for different goals, from minimal images to fully loaded images for testing.
- Modularity
- Any concrete statement (about the software, not the process!) so
people can know what to expect ?
This statement has been that 4.0 would be as 3.10 with minimal changes for relicencing purposes.
3.11 has been proposed as an image with minimal changes, that would be proof of concept for the new process.
a) Readies the image for automated build and testing - i.e. release early and often will be properly possible at last b) Readies the image for our automated/open mantis integration for bug fixes c) Hopefully includes atomic loading for making radical changes to the image possible (i.e. new gui/compiler etc) d) Readies the image for future releases that are assembled by a selection of components, working to a design proposal rather than by an infinite set of incremental changes by one or two control freaks.
(p.s. I need help someone to test and debug (c))
The actual "design" as to what 3.11 results in has been in the form of executable code for over a year, in the ss/Tasks repository. This is probably looking a bit stale now since I have been working on bob.
When was it approved by the community or the elected leadership?
ages ago.
http://installer.pbworks.com/Squeak311Proposal
Was there a decision process involving the community?
http://installer.pbworks.com/Squeak311
Keith