libera/clasp - IRC Chatlog

23:49:08 paulapatience bike: When you get a chance, can you take a look at PR 1572?

23:50:26 bike suuuuure thing. oh the tests pass now, good

23:51:40 bike hm, so we still need the specializations of Integer_O::create even with the auto specializations? oh well. i assume that's due to some recursive include crap

23:53:05 bike yeah you even put a comment explaining that, ok.

23:54:15 bike looks fine to me. merged.

23:54:23 bike thank you.

23:57:32 paulapatience There's just an issue that could crop up in the future depending on how we use Integer_O::create that I mentioned previously (https://irclog.tymoon.eu/libera/%23clasp?around=1714445987#1714445987)

23:59:33 bike how would that be a problem? we're not going to make bignums out of uint32s

0:00:00 paulapatience It's a hypothetical problem, I guess.

0:00:53 bike i'm not too worried about it. ideally we should refactor things so that we can always use the template overloads and always use Integer_O::create, rather than clasp_make_fixnum or Bignum_O::create (outside of Integer_O::create)

0:01:27 paulapatience Ok

0:01:51 bike but that would probably take some untangling of the includes, and fixing really old parts of the code

0:03:08 paulapatience The next thing I will work on when I get around to it is adding more policies for pointer lifetimes, like boost python has

0:03:24 bike very cool.

0:03:58 bike right now i'm working on some compiler stuff that i've really wanted to do for like... at least a year. so i don't want to untangle the clbind const/etc situation right now. but it is on my todo list.

0:04:27 paulapatience It's working fine on my side, so no rush

0:05:13 paulapatience The pointer lifetime policies will simplify my wrapper quite a bit though, so I'm looking forward to doing that

0:06:02 paulapatience I'm presenting my simulations next Monday, so that's pretty cool

0:07:39 bike break a leg!

1:57:42 drmeister Hello.

1:58:03 drmeister paulapatience: Good going!

1:58:31 drmeister zeus is so much faster than the owlsnest machines.

1:58:56 drmeister bike: I moved all the data to zeus and I'm fixing the bugs there.

2:01:37 bike yeah, i mean you're not really supposed to run anything on those login machines, right

2:31:09 drmeister Tell that to the guy running `packmol` for 60 hours.

2:40:53 paulapatience bike: I think something broke recently in Clasp. https://plaster.tymoon.eu/view/4259#4259

2:42:06 paulapatience My code was working before you merged PR 1575

2:42:37 paulapatience Maybe introspect-environment needs to be adjusted, I don't know

2:43:48 bike i don't think it should need adjustment

2:43:54 bike whyyyyy would that slot be unbound. hrm

2:45:02 paulapatience Let me try to get a smaller test case

2:45:58 bike i've got it covered

2:46:18 bike (let ((x nil)) (macrolet ((e (&environment e) (clasp-cltl2:variable-information 'x e))) (e)))

2:46:21 bike but thanks

2:46:32 paulapatience Great

2:47:45 bike hm, might just be a deficiency in getting a cleavir environment from bytecode

3:24:03 bike yeah, the problem is the cleavir-env:variable-info method on bytecode environments. it doesn't put in a type, which it probably ought to

3:24:12 bike lemme see how easy it is to get the actual declared type

3:27:04 bike the answer is not very. so i can put in a dumb fix that just makes all variables type T etc, but i'd rather do it right

3:27:16 bike i'll take another look tomorrow. shouldn't be **too** complicated, knock on wood

3:44:41 paulapatience Ok. I've just checked out an older commit, so there's no rush.

4:56:21 drmeister bike: I finally hit the real bug I've been dealing with in the clustering. Now I can fix it.