freenode/#sbcl - IRC Chatlog
Search
14:18:23
jsjolen
stassats`: But I guess there's some sort of internal segmentation in that big chunk of memory since Linux reports it as such? It's just all contigious now
14:21:16
stassats`
jsjolen: the MMU maps virtual address to real addresses in the big chunk of memory
14:24:10
jsjolen
stassats`: Yeah, that I know :-). The literature I read has shown how to calc. the mapping but described the heap and stack as being discontigious
14:25:41
pkhuong_
each process/thread has an initial stack, but it's just a suggestion on linux. You can point SP anywhere you want. The heap used to refer to the region at the bottom of the address space that grew with (s)brk, but we have mmap now.
19:49:26
pkhuong_
I don't think SBCL is carries enough state across toplevel functions to be capable of superlinear time.
21:36:12
stylewarning
Surprisingly SBCL is way faster than CCL at compiling this huge file containing 100s of generated CFFI bindings
1:43:07
cromachina
can someone explain the purpose of this change in 1.3.18? "minor incompatible change: the result of WRITE-TO-STRING may be a BASE-STRING if all characters written are of type BASE-CHAR."
1:45:34
cromachina
wouldnt this break anything that wants to (declare (type (simple-array character) ...))?
1:49:01
cromachina
do you have an easy way to relax this constraint or force WRITE-TO-STRING to always provide the old string type?