Tim is wildly believed to have written:

> The pain I refer to is that many of the above would make for a cut in
> the backwards compatability story of squeak. Right now we can just about
> run images back to 1.3 or whatever on current VMs. After a change in CM
> format that would no longer be true. 

    I do not see any problems regarding backwards compatibility.

<PLUG>

    Currently MobVM just sniffs at the image then loads the appropriate 
    MegaInterpreter, or YAT (Yet Another Triad, tm ;-)  Interpreter, 
    ObjectMemory, Primitives plugins to run either VI4 or VI3 images.

</PLUG>

> Personally I think that it would be worth it but other people 
> (admittedly demented people lacking in intelligence :-) ) 
> might disagree.

    Yes, I hurrily agree ;-)

    I also think that it's worth it and the time to do it is now. ;-)

    Consequently, I also think that it's worth it to have,
    at the least, MegaInterpreter plugin.

    An ObjectMemory plugin might prove useful as well.

    Cheers,

    PhiHo.


----- Original Message ----- 
From: "Tim Rowledge" <tim at sumeru.stanford.edu>
To: <squeak-dev at lists.squeakfoundation.org>
Sent: Saturday, March 08, 2003 1:26 PM
Subject: Re: Proposal to get to the triad


> Oh, while we're discussing major changes in the organization of images,
> let's not forget the long delayed image format changes. I'd really like
> to try to get it all in one big explosion of alteration so that the pain
> is only experienced the one time.
> 
> At the moment, _my_ list of changes involved here includes:-
> 
> compiledmethod structure change and associated vm alterations
> lose the obsoleteindexed/numbered prim tables from the vm
> cut the internal prim table down to size (we just don't need 7k entries
> anymore)
> assorted bits suggested in the past
> 
> and of course, working out what to do about closures - Anthony's work,
> mixing it with a jitter design etc.
> 
> 
> Other things to consider include
> 
> whether ImageSegments are to be supported in the future or not (are they
> still of any use after the image is super-condensed and packages can be
> added?)
> making headless work across all machines (I see several attempts at
> doing it and it would be nice to have a single way)
> oh, loads of stuff we've argued about.
> 
> The pain I refer to is that many of the above would make for a cut in
> the backwards compatability story of squeak. Right now we can just about
> run images back to 1.3 or whatever on current VMs. After a change in CM
> format that would no longer be true. Personally I think that it would be
> worth it but other people (admittedly demented people lacking in
> intelligence :-) ) might disagree.
> 
> tim
> 
> -- 
> Tim Rowledge, tim at sumeru.stanford.edu, http://sumeru.stanford.edu/tim
> Strange OpCodes: FSM: Fold, Spindle and Mutilate
> 
>