The move to gamma

Doug Way dway at riskmetrics.com
Thu Jul 24 05:24:54 UTC 2003


On Wednesday, July 23, 2003, at 01:56 AM, Daniel Vainsencher wrote:

> We have shortened the beta/gamma stages by quite a bit more than we
> planned. I think before we proceed, we should make sure we are ready.

I think you've identified the main issues.  I've been wondering about 
the possibility of postponing dates myself.

> I see 3 issues that might make this release more buggy than previous
> releases, which I think is unacceptable.
> 1. Have the debugger/context issues stabilized enough to go gamma?

I'm guessing there may be a few more small bugs which show up... 
although I think the most serious problems have been fixed.  Anyone who 
runs across something funny with the debugger, please post a [bug].  
Anthony has been good about looking at these lately.

> 2. Have we fixed all the calls to deprecated stuff made in the image?

Hopefully the senders within the image were fixed when the methods were 
deprecated, although if there are other non-deprecated implementors of 
the same method name, that's trickier.  (we could check the ones which 
don't have other non-deprecated implementors)

The "full" packages outside the image more likely may have problems.

> 3. Do we have a clear idea about how users of this release will be able
> to work at least as well as with the previous scheme of Full releases?
> IOW:
> - is SM1.1 ready so that we can split the current version of each
> package? if not,
> - are all the packages that have been removed in such a state that we
> can pull them in as they are now and thus release a Full image that 
> will
> work?

I've been assuming we would just do the latter.  The packages really 
only need 3.6-final versions anyway.

We do need some testing of these external packages.  For starters, each 
package maintainer could make sure the package loads and passes any 
tests it may have.  (I believe there's a known issue with the PWS 
package, at least.  That should be fixed, or we should consider voting 
on whether to remove PWS from the list of "full" packages, since it's 
mostly obsolete anyway.)

- Doug Way


>
> There may be other issues, those are the ones I could think of. Unless
> we're quite sure that in each of these we have a clear solution that is
> better than 3.5 provides, I think we should delay the move to gamma.
>
> Either way, the time between now and release is quite short, and 3.6
> won't change much before the release. This is a good moment for 
> everyone
> that might still be doing development in something other than the 
> latest
> to upgrade and help with the testing. What we don't catch before the
> release will cause us much more trouble than if we catch it while in
> beta.
>
> Daniel



More information about the Squeak-dev mailing list