Block Closures are here!
Withers, Robert
rwithers at quallaby.com
Wed Jan 23 17:16:13 UTC 2002
> -----Original Message-----
> From: Tim Rowledge [mailto:tim at sumeru.stanford.edu]
> Rob Withers wrote:
>
> > >Looks like a _really_ good start on the SqF Version4 image
> project to
> > >me. Thanks Anthony.
> >
> > So, you don't think we can make it for the 3.2 VM, Tim? It
> would be nice
> > for the SqueakNews issue.
> >
> > Rob
>
> No way; this is a big change that needs some time to settle down. Does
> anyone really think a change like this can be done without bugs? A
> couple have already been fixed and I expect a few more, not to mention
> having a strong feeling that more performance might be squeezable. Oh
> and it might be possible to make changes that make it better
> suited to a
> jitter - better to do it all in one go if possible.
Fair enough. At the least, the alpha VMs and the BC.image could be included
on the CD. I guess I forgot the smilie to indicate that I was (sort of)
joking. It is good to finally see them and they work very well on first
inspection. Even I managed to get a VM built, based off of the SF
code-base. :)
>
> Also there are several other image format changes on the list to look
> into and I haven't had time to read all the code to see which ones are
Yes and there are other VM changes that I would like to see included in the
mainline system - like disabling the class and equivalent bytecodes.
> still open. The worst thing you can do when somebody produces
> a big body
> of work like this is insult it with poor review work.
I think the flurry of activity that has ensued does not represent poor
review work, but rather intense excitement. I wish some of my code
generated 0.01% of the excitement BC has received, but that's the nature of
the evolutionary process, I suppose.
Clearly, there are other issue to resolve. For instance, the patch that
Anthony posted for nested block variables also needs to be fixed in the
CCodeGenerator. There is a method in StarSqueakPlugin that uses nested
blocks that aren't generating local variables for the block arguments. This
means that there is difficulty in generating an interpreter and plugins from
a fully BCed image.
This need for effective review is also why I felt it important to have SUnit
supported in the BC image. (changeset posted yesterday). We should be able
to produce 20-30 tests for BlockClosures, themselves, and 50+ for generators
nad tools that use blocks, like the Compiler, the CCodeGenerator, the
Inspector, and the debugger.
Rob
More information about the Squeak-dev
mailing list
|