On Fri, 2009-06-19 at 01:33 -0400, David T. Lewis wrote:
On Thu, Jun 18, 2009 at 11:39:13AM -0500, Ken Causey wrote:
Please see
http://squeakboard.wordpress.com/2009/06/18/meeting-report-for-6172009/
Thanks Ken,
With respect to:
We spent the bulk of the meeting talking about 3.11 status. We didn't come to any significant conclusions but resolved to increase communication with the release team and compile as much information as possible about the current status of this project and where it is going next.
I would like to request that, in the context of the 3.11 status, the board also discuss the reestablishment of an update stream for Squeak releases. Squeak has been without a usable update stream since Squeak 3.8. In my view, this is simply inexcusable.
I'm sure I must be misunderstanding your intent here. The update stream continues to exist and was used for both 3.9 and 3.10. Of course for the bulk of those two releases most of the updates triggered the download and installation of a mcz (sometimes a mcd), although late in 3.10 use of MC was dropped and standard updates were used again. It has not been used for 3.11 so far because the release team has been concentrating on infrastructure meant to increase automated testing and reduce the effort required to harvest changes. Once that is farther along updates in some form should appear rather quickly, albeit perhaps not in the classic update stream.
<rant> Working on cool new tools to improve the update stream is not a substitute for providing an actual working update stream to the community. Over a period of several months^h^h^h^h^hyears, it has been really quite embarrassing to watch various attempts to distribute updates in a frankenstream of Monticello archives and 10MB monolithic images. This is followed by complaints that nobody is paying any attention to the updates. Finally, we have great distress and wringing of hands when it is discovered that our change histories have been destroyed (because there was no update stream, d'oh!) and we can't figure out our licensing because we don't even know who made changes, or how, when and why the changes got into the image. </rant>
Regarding the perceived 'distress': there is no less information regarding contributors in a MCZ than in a fileout. It just adds another location that has to be checked and is just not quite as easy to examine as a small fileout.
I think you are misunderstanding the difficulties of the licensing problems, it has absolutely nothing to do with whether a change was submitted as a fileout or as an mcz. We have the same information in either case, perhaps even more in mcz. At this point the main problem is just trying to figure out what amount of change to a method, for example, by a person is significant enough to warrant a rewrite if that person has not submitted a license agreement and, if a rewrite is needed, what procedure needs to be used for that. Discussion with the Software Freedom Consortium is going on but frankly the number of people both willing to help with this and sufficiently knowledgeable of the situation is so small that progress is exceedingly slow.
Dave
Ken