I think the current situation is somewhat unusual, in that a major 
update (the String refactoring) was added to a x.y beta/gamma release, 
but not added to the following x.y+1 alpha release.

In this case I think it was warranted... basically there were already 
some major changes to String made early in 3.8alpha (because of m17n), 
so if we're going to revamp those changes, we should really try to do 
it in the same release if at all possible.  (So we don't have 3.7 with 
old-style Strings, 3.8 with major String changes, and 3.9 with yet more 
major String changes.)  Even if this means delaying 3.8 a bit.

And also, there will be 3.8.1, 3.8.2 etc releases which can contain bug 
fixes.

Actually, I don't think 3.9alpha had really diverged that much from 
3.8gamma (before the String changes)... the only major change was 
Diego's look changes, but a lot of that was just image/preference 
changes.  So I don't think it will be too hard to port the String 
changes forward from 3.8gamma to 3.9alpha.  I guess the idea is to 
hammer out the (String change) problems in the 3.8gamma version before 
porting it forward to 3.9alpha.

So in summary, I don't think this will be a particularly common 
situation.

- Doug


On Apr 15, 2005, at 2:25 PM, Ken Causey wrote:

> As Bert's chart on http://minnow.cc.gatech.edu/squeak/275 shows the
> sequence of updates between for 3.8 and 3.9 is quite complicated.  At
> times (including now) we've had a situation where there are updates in
> the 3.8 update stream that are not in the 3.9 update stream.  Now maybe
> I just haven't been paying much attention in the past but this is not a
> common occurrence in the history of Squeak development I don't believe.
> Right now it is causing us in the Janitors team a bit of a headache as
> it's a bit of a toss up at times to what image a given fix may or may
> not apply.  We've had a policy of testing everything against a fully
> updated 3.9 image and this has been fine in general but is right now a
> problem with all the String updates that are in 3.8 but not in 3.9.
>
> What I'm wondering is whether we expect this to be a more common
> situation in the future than it has been in the past.  To be more
> explicit do we expect it to be not uncommon in the future for there to
> be a situation in which y.x has some number of updates that y.(x+1) 
> does
> not?  If so then we (Janitors and friends) will need to plan for this I
> think.
>
> Ken
>