libera/#clasp - IRC Chatlog
Search
17:38:05
Bike
now there's the question of how to do the attributes, which involves comparing strings
18:17:35
yitzi
Bike: drmeister told me that snapshots were broken on Mac. Did you know anything about that? They were broken on Arch last month...but now appear fixed.
18:27:12
Bike
we had someone running into weird segfaults on their mac, but i don't know if that's what drmeister is talking about
18:44:16
drmeister
With an environment variable I have to shut down emacs, set the environment variable and then start slime within the emacs.
18:44:33
yitzi
Hello....snapshots are working for me on Linux again...do you recall an error message from Mac?
18:44:57
drmeister
` /usr/local/Cellar/cando/2.2.0-64-g3a6244cc4-g1158b1a8/share/clasp/src/gctools/snapshotSaveLoad.cc:3627:snapshot_load The newCodeClient literals range 0xa8 - 0xb0 must be within the bounds of the newCodeClient data range 0x7fc99f036f40 - 0x7fc99f74ce20 and it is not`
18:52:36
drmeister
Yes, the snapshot save/load code does a lot of pointer arithmetic - somethings wrong somewhere.
19:07:29
drmeister
It will crash Rosetta2 if I let it sit like this for a while. It did that yesterday.
19:23:41
yitzi
Ok, I've got it here as well. I didn't see it in a development clasp. I'll try building cando to check that.
19:35:21
yitzi
Hmmm...we are only building bottles for Big Sur and Monterey. I don't think GitHub supports ventura yet.
19:41:58
yitzi
Could be. I think homebrew might have their own image for macos-13 which is why they are building for Ventura.
20:07:00
yitzi
There are automated kernel installation stuff in clj and cando-jupyter. I don't know if the kernel spec file he gave has something custom about it though.
20:14:31
yitzi
drmeister: The issue has nothing to do with the Monterey bottle. I've reproduced the problem with a development cando. save-lisp-and-die works from a fresh cando instance, but the error happens when using it from inside cando-user-install system.
20:15:25
Bike
huh, i'm getting that stupid cst-read error. didn't we fix that? this is a freshly cloned cando, too
20:22:21
Bike
nope, still getting cst-read. it looks like this is while trying to compile cl-jupyter. but i have :jupyter t in my config, so that should be pulled based on repos.sexp, and that's got to be up to date, right?
20:24:41
yitzi
I don't think you need :jupyter t ... drmeister doesn't use that when he does the custom kernels. I
20:28:05
yitzi
Oh, and when you use :jupyter t there is a quicklisp copy inside the dependencies folder which probably needs quickclasp updates also.
20:28:54
yitzi
So definitely turn it off. That could be related to the issues you were having with quicklisp before.
20:43:46
Bike
still doin it. now how do i even go about figuring out where the old cl-jupyter is lurking
20:46:36
Bike
looks like it's loading from a copy of cl-jupyter in my dev directory. i probably put it in quicklisp local projects ages ago
20:47:24
Bike
or maybe not, but whatever, it's trying that stupid directory for whatever reason, so goodbye to that