Smalltalk › Squeak › Squeak VM

with permission I would like to claim slots 35 & 37 in the special objects array...

_‹ Previous Topic Next Topic _›

Classic

List

Threaded

5 messages Options

Eliot Miranda-2

with permission I would like to claim slots 35 & 37 in the special objects array...

Hi All,

these two slots are the unused MethodContext and BlockContext prototypes. They're either side of class BlockClosure. I'd like to use one of them for threaded callbacks (the foreign callback process slot, upon which a callback from a foreign thread can be run) and keep another in reserve. In Smalltalk these are indices 36 & 38 in SmalltalkImage>>recreateSpecialObjectsArray. Objections, encouragements, alternatives?

best

Eliot

johnmci

Re: with permission I would like to claim slots 35 & 37 in the special objects array...

Sure no problems here

On 2010-08-26, at 1:31 PM, Eliot Miranda wrote:

Hi All,

these two slots are the unused MethodContext and BlockContext prototypes. They're either side of class BlockClosure. I'd like to use one of them for threaded callbacks (the foreign callback process slot, upon which a callback from a foreign thread can be run) and keep another in reserve. In Smalltalk these are indices 36 & 38 in SmalltalkImage>>recreateSpecialObjectsArray. Objections, encouragements, alternatives?

best
Eliot

===========================================================================

John M. McIntosh <[hidden email]> Twitter: squeaker68882

Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com

===========================================================================

Andreas.Raab

Re: with permission I would like to claim slots 35 & 37 in the special objects array...

In reply to this post by Eliot Miranda-2

Do we really need to recycle these slots? Why not just appending these
to the end?

Cheers,
- Andreas

On 8/26/2010 1:31 PM, Eliot Miranda wrote:

>
>
>
>
> Hi All,
>
> these two slots are the unused MethodContext and BlockContext
> prototypes. They're either side of class BlockClosure. I'd like to use
> one of them for threaded callbacks (the foreign callback process slot,
> upon which a callback from a foreign thread can be run) and keep another
> in reserve. In Smalltalk these are indices 36 & 38 in
> SmalltalkImage>>recreateSpecialObjectsArray. Objections,
> encouragements, alternatives?
>
> best
> Eliot

David T. Lewis

Re: with permission I would like to claim slots 35 & 37 in the special objects array...

No objections from my point of view, although I do slightly prefer
Andreas' approach.

These two slots were present in Squeak 3.6 but were apparently unused.
I presume that they were intended for the (now dormant) NewCompiler
work. In the interest of general tidiness it would be best to recycle
the slots, but for the sake of courtesy to anyone who still has an
interest in NewCompiler it may be better to retain them a while longer
(I think that there may be some interest it in for research purposes,
even if it does not provide practical benefits otherwise).

I note also that the size of the special objects array can serve
as a de facto version identifier, so it does not really hurt to
have the array grow a little bit when some significant new feature
is added.

Dave

On Thu, Aug 26, 2010 at 01:51:16PM -0700, Andreas Raab wrote:

>
> Do we really need to recycle these slots? Why not just appending these
> to the end?
>
> Cheers,
> - Andreas
>
> On 8/26/2010 1:31 PM, Eliot Miranda wrote:
> >
> >
> >
> >
> >Hi All,
> >
> > these two slots are the unused MethodContext and BlockContext
> >prototypes. They're either side of class BlockClosure. I'd like to use
> >one of them for threaded callbacks (the foreign callback process slot,
> >upon which a callback from a foreign thread can be run) and keep another
> >in reserve. In Smalltalk these are indices 36 & 38 in
> >SmalltalkImage>>recreateSpecialObjectsArray. Objections,
> >encouragements, alternatives?
> >
> >best
> >Eliot

stephane ducasse-2

Re: with permission I would like to claim slots 35 & 37 in the special objects array...

On Aug 26, 2010, at 11:58 PM, David T. Lewis wrote:

>
> No objections from my point of view, although I do slightly prefer
> Andreas' approach.
>
> These two slots were present in Squeak 3.6 but were apparently unused.
> I presume that they were intended for the (now dormant) NewCompiler
> work.

Thanks david for the cautious words. The new compiler is far from dormant in fact.
Jorge worked a year on it. Since last week new compiler is called
Opal (because this is the see around here where we can see england).
Marcus, jb, jorge are working a lot on it. Marcus should send some announces about they progress.
There is also a mailing-list and the result is quite pleasant. The goal is to use it in Pharo1.3.
So far large clean up to IR, decompiler working for fullclosures.....

> In the interest of general tidiness it would be best to recycle
> the slots, but for the sake of courtesy to anyone who still has an
> interest in NewCompiler it may be better to retain them a while longer
> (I think that there may be some interest it in for research purposes,
> even if it does not provide practical benefits otherwise).
>
> I note also that the size of the special objects array can serve
> as a de facto version identifier, so it does not really hurt to
> have the array grow a little bit when some significant new feature
> is added.
>
> Dave
>
> On Thu, Aug 26, 2010 at 01:51:16PM -0700, Andreas Raab wrote:
>>
>> Do we really need to recycle these slots? Why not just appending these
>> to the end?
>>
>> Cheers,
>> - Andreas
>>
>> On 8/26/2010 1:31 PM, Eliot Miranda wrote:
>>>
>>>
>>>
>>>
>>> Hi All,
>>>
>>> these two slots are the unused MethodContext and BlockContext
>>> prototypes. They're either side of class BlockClosure. I'd like to use
>>> one of them for threaded callbacks (the foreign callback process slot,
>>> upon which a callback from a foreign thread can be run) and keep another
>>> in reserve. In Smalltalk these are indices 36 & 38 in
>>> SmalltalkImage>>recreateSpecialObjectsArray. Objections,
>>> encouragements, alternatives?
>>>
>>> best
>>> Eliot