Thanks for your confidence. Since one of the goals for Parrot seemed to be the ability to target multiple languages to a single backend, I figured that we had to have at least two languages to work with. That, and Scheme seemed to be a fairly simple language to parse.
My code is currently in CVS, and I'd certainly be willing to maintain the archive but I'll defer to the wishes of the list. I'm hoping I can add support for both real and integer types this weekend, but various offline tasks have been occupying my attention. It's also helping me coalesce some ideas WRT memory support. One simple implementation that I've got code for here at home is a 'set_i_m' and 'set_m_i' instruction that copies an integer register into a memory cell, which will survive no matter what the state of the integer register stack is. There are, of course, many problems and issues with this implementation, but I'm afraid to manipulate the string or PMC stack to create the same effect. The string stack would cause a great deal of overhead and pervert the notion of 'string storage' into 'string and arbitrary data storage', and PMCs aren't implemented and the notion of executing register warps on such a large data structure to implement such things as arrays strikes me as terribly inefficient. -Jeff <[EMAIL PROTECTED]> "Gregor N. Purdy" wrote: > All -- > > > Judging by the patches, this was a much earlier version than I intended to > > post. In the current version, min and max are now implemented, and test files > > evaluate two-operand and three-operand versions. The (<>=) operands work now > > with more than two operands, and I've added tests for these with three > > operands. > > > > Answering questions: Operators should now use no more than two registers at > > any given time. The bug where _save returned an unused temporary has been > > fixed, > > > > As to whether we can get it into languages/scheme, I'd be pleased to add it > > but I don't have CVS access. I'm also working with a patch to introduce a > > 'set' instruction that writes to and from a temporary memory cell. > > I support having this in CVS. The more languages the better. It does need > a maintainer with CVS access though one way or the other. I'd like to see > the folks with other language implementations speak up again about their > current status and desires to have their stuff in CVS. I think having > these languages available for people to play with helps us keep the > interest in and usability of Parrot high. > > Regards, > > -- Gregor
