freenode/#clasp - IRC Chatlog
Search
4:14:08
Bike
it looks like global value numbering is also supposed to work with operations, like if x = y then (car x) = (car y), which seems a good deal trickier
4:15:45
beach
But it could be worthwhile. I often write things like (car x) several times rather than introducing a temporary.
4:16:27
beach
Again, it gets tricky in the presence of threads and other complex control structures.
4:17:59
Bike
well, for a start i can ignore cells, which makes it easier. i'm still not sure what to do with cells in general
14:36:09
drmeister
The problem I fixed last night I found by listening to the compiliation warnings.
14:39:53
drmeister
I'm noticing something weird though. It's not clear yet, but when I run clasp under the debugger it's like linking doesn't work properly. I get .fasl files with zero length and that's a show stopper.
14:41:14
drmeister
I can't even fathom why that would make a difference. If you do try to run clasp when it is connected to the debugger and you see that - can you confirm it?
14:42:55
drmeister
Let me put it this way - if you are tempted to run under the debugger because clasp backtraces are nonexistant and you start seeing fasl files with zero length - don't panic - I've seen the same thing. When I don't run under the debugger - I haven't seen it.
15:58:57
drmeister
Clasp is working pretty well - cl-jupyter builds and runs - that means Clasp compiles and compiles a whole mess quicklisp systems.
15:59:39
drmeister
I want to get up to the latest sicl - this means figuring out what is going on with that cc_writeCell call that is failing.
16:01:05
drmeister
If we could get a framework for adding inlined code for AREF then I can keep adding to it.
16:03:22
drmeister
What I'm thinking is if you could help me get started on the inlining and code generation stuff then I can keep working on that alongside other things that I need to implement.
16:03:57
Bike
like, mapcar and friends can be inlined and that's not even anything to do with cleavir, for one
16:04:32
drmeister
Meanwhile you could continue working on kildall and other compiler optimization while taking a look at the cl-jupyter notebooks and jupyter widgets stuff.
16:05:17
drmeister
Kevin and Hana will be working on that full time - but they don't know much Common Lisp. So if you could help them there that would be great.
16:05:33
drmeister
I think we can get some starter code put together for them and let them expand on it.
16:06:29
drmeister
I have a rough idea of what is involved to implement widgets in Common Lisp. I know where the Python whos functionality we are copying is and I have hacked it to dump logging information of the communication that we can work to replicate.
16:07:27
drmeister
mapcar can be inlined? Neat. I recently fixed them so that they wouldn't CONS - inlining would be even better.
16:08:17
drmeister
Right. My old implementation was before I figured out how to do more dynamic stuff on the C++ stack.
16:08:20
Bike
yeah, that sounds fine. i like teaching. and rapidly learning jupyter while having to explain it sounds like a good way to make it solid
16:09:58
Bike
just basically (mapcar #'foo bar baz) = (loop for x in bar for y in baz collect (funcall #'foo x y)).
16:12:23
Bike
and if loop doesn't work you can just write it out with dotimes or tagbody or whatever. loop's just the intuitive way to do it.
16:13:15
drmeister
It should be fine - mapcar is only used three times in loop2.lisp and not in any way that would cause problems.