New VM in progress
Tim Rowledge
tim at sumeru.stanford.edu
Thu Dec 13 01:04:12 UTC 2001
Scott A Crosby <crosby at qwes.math.cmu.edu> is widely believed to have written:
> What are the new enhancements that a new incompatible release would offer?
> Maybe we should make a full list. Showing people the features of the new
> (incompatible) VM might make it easier to get it out of the door.
>
> * Any other speed patches (other than mine) Yes? No?
Not sure any particular speed improvements are incompatible-making
> * MethodClosures Yes? No?
I sure hope so. Sounds link anthony has this well in hand
> * Not paying a fullGC on every 'openFile'. Yes? No? Platform dependent?
Don't think that is vm dependent, nor incompatible-making
> * faster become with forwarding pointers Yes? No?
Not yet convinced it is really faster. Data would be nice to see on this
one.
> * Instance based prototyping patches Yes? No?
Hmm. Not a fan of instance based, but Rob Withers and I got a simple
demo hack for something like this working at oopsla this year. Well,
delegation, anyway. Or something. Rob? Stop me before gibber too
much...
> * Easier (not-incomatible) integration of Jitter? Yes? No?
Err, not sure I understand. How is jitter incompatible right now? Last
time I checked (ages ago, admittedly) it was a plugin that just took
over if it was there. I'd like to make the 'normal' execution stuff be a
plugin so that it could be dropped if you _do_ have a jitter.
> * More flexible lookup strategies (Mirror, etc) Yes? No?
See note above about stuff Rob & I did.
> * Switch to VMMAKER (is this the definite future?)
I hope so after all the &$%^ing work it took. Just trying to make the
exception handling play nice almost melted my brain.
>
> What have I omitted?
Oh, how about getting the prim number bits de-separated in the method
header, getting rid of a fairly large number of back-compat prim numbers
in favour of the named stuff, removing some of the ITIMER related
stuff... probably more.
tim
--
Tim Rowledge, tim at sumeru.stanford.edu, http://sumeru.stanford.edu/tim
Strange OpCodes: MW: Multiply Work
More information about the Squeak-dev
mailing list
|