freenode/#sicl - IRC Chatlog
Search
4:18:09
beach
I think today I will continue with the work of loading the prerequisites for the sequence functions. Yesterday, I was able to load the SICL-UTILITIES system, but the sequence functions use a bunch of macros that need to be loaded before the code for the first sequence function can be defined.
6:20:20
Harag
Well the good news is I got SICL up and running with little effort, and in lest time than I thought, after refreshing all the required repo's it took about 5 minutes of compiling etc and I had a sicl repl
6:29:40
beach
Yes, currently bootstrapping takes around 5 minutes. But code generation is currently turned off, so that saves more than half the bootstrapping time. And there are some big chunks missing, like Eclector and the compiler, currently executed by the host. On the other hand, heisig is working to improve the performance of the bootstrapping code.
6:31:08
beach
Now, it occurred to me, that there are lots of things that you don't need. Most of the bootstrapping procedure exists in order to create metaobjects represented in the particular SICL way with a header, a rack, etc. You should just use the host version of all that, I think.
6:31:38
beach
But I haven't really given any thought to the details of how to do that, since it is not the main mission with SICL.
7:16:29
beach
The sequence functions that heisig wrote depend on his library "Fast Generic Functions", and I will need that one for the generic arithmetic functions as well, so I think I'll start by trying to understand that library. I also need to determine whether the existing code can be used in SICL directly, or whether some adaptation is required.
7:20:25
beach
So I will determine whether that adaptation can wait or whether it needs to be done up front.
7:20:39
splittist
It would be pretty incredible if a cross-compiler library could also act as infrastructure of a compiler itself.
9:27:16
heisig
beach: I am very much interested in hearing your thoughts on the fast-generic-functions library.
9:28:42
beach
I may have to wing it for now, i.e., make the protocol(s) work but without the performance benefits.
9:29:38
beach
That's true. But I still need to figure out what SICL support is needed so that the protocol is respected.
9:30:38
beach
I should also look into how you use trivial-macroexpand-all. It may not be needed for SICL since we control the compiler.