Linker is busy for ten hours now. It just cache-trashes, I hope I can bootstrap soon, sick of that.
I am writing on the document generator, and thought a bit more about how to write it right, combinator style, which was my original approach to this language. I'll refactor after the bootstrap somewhat, some things I just didn't do right.
My laptop doesn't even seem to want to spend time on it anymore, it just memory hogs and sits there, without any reasonable progress or actual processor activity. I could code more, which is nonsense, or just upgrade from 2 to 8 GB. Which, actually, makes the most sense. The C compiler is good, no sense in going back for just an upgrade and a wait, it's just getting rid of ML...
(A small analysis. I can see the process jumping between 1 and 2 GB each half hour or so. As the AST is translated to lambda terms, to combinator terms, to code -where it halts,- they grow in size. Since I still only have simple transducers, the growth in size is linear. It stalls on the renaming of variables in the pseudo assembly code. I wrote the collection of free variables as a simple bottom-up procedure collecting and merging lists of them, which, in the end, is just damned expensive.)
No comments:
Post a Comment