Search
Monday, 12th of November 2018, 11:09:45 UTC
11:09:50
flip214
thanks for all the help, could figure it out
11:41:24
pfdietz
No failures with nearly a day of testing on eight cores.
11:41:57
pfdietz
(well, four cores, doubled with HT)
13:23:22
stassats
flip214: just hold on, the fact that you had ask here means we're doing something wrong
13:24:02
stassats
ok, [error printing frame], is the whole frame unprintable? i bet the function name is printable
13:29:45
flip214
stassats: I'm sorry if I seem to blame someone around here. That's not my intention!
13:29:57
flip214
I was asking because an older version of sbcl did work fine.
13:30:04
stassats
that's not what i meant
13:30:32
flip214
the causing change was 14c8a92609, and that a bad workaround from me
13:31:52
flip214
and *then a bad workaround
13:31:56
stassats
i meant sbcl should present the errors in a way that doesn't require #sbcl to decipher
13:32:46
flip214
the swank stack was broken - in a part that modified _internal_ sbcl thingies
13:32:55
flip214
see https://github.com/slime/slime/issues/213
13:34:09
flip214
reminded me of "The night watch" ( I HAVE NO TOOLS BECAUSE I'VE DESTROYED MY TOOLS WITH MY TOOLS.
13:34:32
stassats
i can see just one thing i can improve so far, [error printing frame]
13:34:53
flip214
https://www.usenix.org/system/files/1311_05-08_mickens.pdf
13:35:33
flip214
stassats: well, slime does (setf (fdefinition 'sb-impl::%print-unreadable-object) ...)
13:36:10
flip214
if something in there is broken (or gets broken _at_that_level_), SBCL won't be able to help any more.
13:36:25
stassats
maybe it does, doesn't concern me just yet
13:36:28
flip214
anything you can do to make that easier to debug next time would be very welcome, of course!
13:37:05
stassats
and it's just for swank-presentation-streams.lisp, just don't use it
13:38:48
flip214
well, vlime uses that, so I can't easily skip that.
13:40:16
stassats
ok, sbcl already has an option for this, slime just doesn't use it
13:41:47
stassats
https://gist.github.com/stassats/8550c663d7bfc8fc3243b6fe9e5dd2f0
13:42:34
flip214
stassats: that is much better, would you append issue 213 or tell me what to note there?
13:42:58
stassats
nah, i'll just fix it myself
13:46:58
stassats
flip214: ok, pushed that
13:53:34
stassats
now a bit of an improvement to sbcl itself
13:53:39
stassats
(STRINGIFY-OBJECT #<error printing BAR {100D304273}>)
14:09:58
pfdietz
20M test cases through the random tester, no failures. stack analyze seems to be working.
14:14:23
flip214
hm, SF git is still at 4a37e8676
14:16:25
flip214
minion: memo for stassats: how about doing https://sourceforge.net/p/sbcl/mailman/message/33102884/ so that the interface becomes defined?
14:16:25
minion
Remembered. I'll tell stassats when he/she/it next speaks.
15:02:41
stassats
flip214: SF git doesn't have slime
15:02:41
minion
stassats, memo from flip214: how about doing https://sourceforge.net/p/sbcl/mailman/message/33102884/ so that the interface becomes defined?
15:04:21
stassats
flip214: nah, only if somebody else does that
15:34:28
scymtym
mcclim is also interested
15:47:35
flip214
stassats: ah, didn't know you pushed to slime. I thought sbcl. ack, got it!
Monday, 12th of November 2018, 23:09:45 UTC