libera/#climacs - IRC Chatlog
Search
9:10:43
beach
Shouldn't there rather be a :SLOT-NAME kind just as there is a :FUNCTION-NAME and a :TYPE-NAME?
9:28:43
beach
Well, I decided on a mixed approach to making progress. I started working on a version of CST-to-AST that uses Iconoclast. And I designed it to be an alternative to Cleavir/CST-to-AST used by SICL now, just for the convenience of being able to use a pre-filled SICL environment.
9:31:59
beach
I think I understand how to structure CST-to-AST. The builder will contain both a client and an environment as hinted by scymtym. I take apart the builder in order to call the main entry point of CST-to-AST (called CONVERT), and I create a builder from a client and an environment in order to call a parser.
10:54:12
beach
Also, for slot options, there does not seem to be a node for the individual options, like for :INITARG and :INITFORM. That means the source position for the keywords can not be obtained, right?
10:55:05
beach
There is a node for the initarg name, and for the initform form, but none for the pairs.
12:11:42
scymtym
beach: i think i used VARIABLE-NAME as an approximation when i hadn't fully decided whether to use proper nodes for all names. so you are right, it should be SLOT-NAME
12:13:10
scymtym
by "adding a parser for a standard macro" do you mean an ast class and builder methods or the actual parser? in the latter case we would have to coordinate
12:17:40
beach
Sounds good. Like I said, I am doing this incrementally and it is not on any critical path.