At 10:53 PM 8/12/2007, Jeff Read wrote:
Marc,
On 8/12/07, Marc Feeley feeley@iro.umontreal.ca wrote:
Some of you wanted to know why I am voting against ratification of the R6RS. Here are the details.
You said it much better than I ever could.
Certainly much better than I did, but I'm not (yet) an implementer, and my targets included the "minimalists", trying to get them to buy into a proper core (which R5.97RS most certainly is not) that includes an acceptable library system, and a set of official standard libraries that would include reference implementations, so the floor of a compliant although not very fast system is not too high, and we can hopefully enjoy the best of both worlds.
Don't know if the above is really practical, but it was discussed on comp.lang.scheme and seems to be the best way to try to square this circle. I take it as a given that the "failure of the community to build a set of portable libraries" (to paraphrase one voter cited in my essay) needs to be addressed (properly!), and however we might e.g. like Snow, it's not getting buy-in from some comments people made....
But if I help to SMP Gambit, I sure don't want the core to be polluted by all that R5.97RS adds.
That's the one tasty area where I want to do implementation work, although Marc's analysis of what it would require and the performance hit it would entail for Gambit is ... sobering. But I'd like to use Gambit in a multi-core world beyond the Termite NUMA model.
- Harold