[Squeakfoundation]3.5 release timing (was Re: Outstanding 3.4 bugs?)

[Doug Way]

On Sunday, March 9, 2003, at 06:38 PM, Tim Rowledge wrote:

> Doug Way <dway at riskmetrics.com> wrote:
>
>> So, unless there are major objections, we will go with the 1-month
>> bugfix plan for 3.5.
> I'll just raise again my objections to the idea that it is possible to
> do any sensible beta in only a couple of weeks and any plausible gamma
> ditto. If anyone tries to incorporate any new work whatsoever it will
> completely blow the faintest possibility of success.

Well, to be honest I am still somewhat on the fence about this.  (as 
far as a 1-month bugfix release being *really* necessary)

But there are some advantages to doing this and I think it can be 
workable.  We would need to put some limits on the number and kind of 
bugfixes that are accepted... perhaps no more than a dozen or so of the 
most important fixes, and they must all be small in size.  Of course 
this doesn't guarantee anything, but allowing more & larger fixes will 
increase the risk.

A 3-week beta+gamma is somewhat short, but not ridiculously so for a 
Squeak release. ;-)  I believe the Squeak releases prior to 2.8 had 
(unofficial) beta stages which were typically about this long.  Which 
is not to say that that was good software-release practice either, but 
at least in this case there will be less risk than in the typical 
release in which hundreds of changes (some large) go in.

Arguments against a 1-month bugfix release are basically that while we 
do have two much-discussed "serious" bugs in 3.4 (the project-saving 
bug & the class compilation bug), the former still doesn't have a 
reliable fix yet, and the latter is arguably not particularly critical. 
  The rest of the bugs being fixed recently seem to be the 
garden-variety bugs present in every Squeak release.  Also, I don't see 
any reason to postpone proceeding with removing packages from the 
image.  And IMHO bug-fixing could move at a more unrestrained pace 
during a normal release as opposed to a shortened release.

Anyway, we can further discuss arguments for/against the 1-month plan, 
and I could probably be convinced either way, but I'd still like to 
announce some sort of 3.5 release plan within the next couple of days 
on squeak-dev (as mentioned in my last message).

- Doug Way