libera/#sicl - IRC Chatlog
Search
15:50:46
mfiano
Right now I'm thinking more about program modules in my design. I am questioning their granularity, and the criteria on which they should become external libraries.
15:54:49
beach
mfiano: You are excused. I think it requires intimate knowledge about SICL bootstrapping and compilation.
16:15:04
Bike
beach: where are you using ctype? i'm writing a more comprehensive manual and contemplating interface changes and don't want to break things
16:15:50
beach
ctype is the native implementation for SUBTYPEP in SICL, and I am using it in COERCE.
16:28:37
beach
I am off to fix dinner for my (admittedly small) family, and then spend time with her. I'll be back tomorrow.
16:30:18
Bike
"Literals that are present in source code are already present in the instruction stream in the untied code object" i thought that, essentially, things like literal symbols would be treated as load-time-value forms, but this is written like only actual load-time-value forms are treated as such and everything else is more immediate
3:03:25
beach
Bike: Thanks! There was a time when the FASL was meant to contain only native instructions, so we had to deconstruct every literal. Now that the FASL contains a printed version of the AST, if the reader can read a literal in the FASL, there is no need to deconstruct it.
3:04:56
beach
And there was a time when literals had to be created by the top-level function resulting from the compilation of a FASL, and then transmitted as shared lexical variables to inner functions. Now, the literal is an immediate value in native code.
3:10:08
beach
This new solution means we have the same advantage as a Common Lisp implementation that goes to great length to define NIL as a fixed address, except we have that advantage for every literal.