| timo | no valgrind on macos | 01:02 | |
| [Coke] | try leaks (man leaks) | 01:11 | |
| build is btroken for me on osx right now. :( | 01:19 | ||
| oh, wait, I left bad debug code in, my bad. :) | 01:21 | ||
| before the build, the say command died. I built head in local dir... now neither local nor installed die. | 01:25 | ||
| here's the leaks output anyway: gist.github.com/coke/fc158b257153c...b292646f68 | 02:18 | ||
| that is NOT a debug output, if it'd be helpful, can rebuild and rerun | 02:19 | ||
| timo | the crash happens in the "repl" script of the Repl module when using Terminal::LineEditor, just doing the "say" in rakudo itself doesn't do anything. the explosion happens in the getattr (MakeRaw) in enter-raw-mode (Terminal::LineEditor::something) | 02:23 | |
| but the explosion there is just a "malloc has found out that the datastructures on the heap have become corrupted at some point" | |||
| well, it's calloc that figures it out, the calloc when creating the Termios CStruct instance | 02:24 | ||
|
05:05
japhb left
05:09
japhb joined
07:54
kjp left
07:55
kjp joined
|
|||
| timo | ... the Termios struct definition is too small ... | 09:16 | |
|
09:35
MasterDuke left
09:59
sena_kun joined
|
|||
| [Coke] | the freshbuilt raku doesn't explode in the REPL - the old one does. | 17:54 | |
| (leaks doesn't like working with the older rakubrew version of raku - keeps segfaulting) | 17:57 | ||
| no, now it's just segfaulting all the time, wtf. | 17:59 | ||
| timo | not mimalloc, though right? | 18:31 | |
| [Coke] | that wasn't in the list before, no - also I was calling leaks incorrectly just now. :| Sorry for all the noise. | 18:47 | |
| timo | the list? | 18:51 | |
| [Coke] | gist.github.com/coke/fc158b257153c...b292646f68 | 18:54 | |
| gist.github.com/coke/f6df56e5efc1f...44faf1c705 - here's one with the repl variant. | 18:55 | ||
| (looks like they are all calloc) | 18:56 | ||
| Use of uninitialized value $args{"os"} in string eq at Configure.pl line 469. in the config in moarvm | 18:59 | ||
| timo | mimalloc is a different allocator library | 19:12 | |
| so not sure if leaks can handle that, or if it depends on internals of the malloc impl in question | |||
| [Coke] | dug into current status of valgrind on m1 - "so close" | 21:42 | |
| (github.com/LouisBrunner/valgrind-m...issues/56) | 21:43 | ||
|
22:29
sena_kun left
|
|||