[VM-dev] Names versus attribute numbers (was: Re: adding external plugins to Squeak 4.2 on the Mac)

David T. Lewis lewis at mail.msen.com
Wed Jul 20 15:37:30 UTC 2011

On Wed, Jul 20, 2011 at 05:11:00PM +0200, Bert Freudenberg wrote:
> On 20.07.2011, at 04:40, David T. Lewis wrote:
> > So in my view it is a good thing to use explicit names, even
> > if it does have some technical disadvantages.
> > 
> > Dave
> Should the primitives take symbolic names too, then? They could easily be changed to accept a string in addition to an index.

I don't really think that this matters one way or the other, as long
as it is consistent from the point of view of the image. The point
I want to make is that numbered system attributes or vm parameters
imply coordinating the definition of these parameters amongst the
various developers and VM projects (i.e. parameter number N should mean
more or less the same thing on RoarVM as it means on Cog, etc). In
contrast, it is unlikely that any two VM development projects would
separately define a primitive called #primitiveNumberOfActiveCores,
and therefore this approach requires less coordination between
the various interested parties. It is also simpler from the point
of view of the image, because it does not require the image to
interpret some parameter value based on what flavor of VM it happens
to be running on.

As a technical matter I don't much care how these things are
implemented. As a social and process issue, I would prefer
to avoid things that require coordination and agreement between
the interested parties, because these things tend to inhibit progress
for the various VM projects (RoarVM, Cog, 64-bit images, etc) and
also tend to produce confusion between branches (a personal
concern for me).


More information about the Vm-dev mailing list