Search
Wednesday, 22nd of November 2017, 8:34:41 UTC
14:34:41
stassats
yak shaved a single change to 7 commits
18:29:52
|3b|
did windows sbcl change how it decides where (user-homedir-pathname) is recently?
18:33:26
stassats
actually, i don't know
18:33:51
stassats
i didn't change it, but now scymtym has been doing some windows stuff
18:34:42
stassats
|3b|: 2fc73e45c983c6f77775d61c174322b5642d3d0a 3f30bba41bc66e5bd9ace84446cecac2db8b4d44
18:34:53
stassats
|3b|: what did it break for you?
18:36:14
|3b|
hmm, maybe something else broke, gitk isn't working properly either
18:36:34
|3b|
getting /Users/foo instead of msys homedir
18:37:49
|3b|
gitk working again after restarting
18:43:19
|3b|
yeah, looks like it used to check HOME first, now just goes straight to USERPROFILE
18:44:19
stassats
right, will put it back
18:47:18
|3b|
also, condition.impure.lisp seems to be hanging at :printable-conditions
18:47:34
stassats
possible, i don't run the tests on windows
18:48:31
stassats
|3b|: HOME should be back in
18:48:35
|3b|
yeah, reasonable choice :)
18:49:17
stassats
why on earth would it fail
18:58:10
|3b|
rebuilt to get HOME fix, and now it doesn't hang on :printable-conditions
18:59:08
stassats
i haven't rebuilt yet, let me see
18:59:17
stassats
that would be silly, though
19:00:31
|3b|
and now that i think about it, my sbcl is patched... is that test likely to involve backtraces at all?
19:00:37
stassats
it doesn't fail in powershell...
19:00:50
stassats
the place where i can get backtraces on interrupt
19:01:35
|3b|
fixing home means my default coding system is utf8 now, that could affect printing things
19:01:53
|3b|
though i guess tests probably ignore sbclrc?
19:02:16
stassats
they should, but there's no encoding involved there
19:03:54
stassats
the encodings are different between cmd and mintty
19:08:34
stassats
right, not dependent on the encoding
19:08:58
stassats
wait, i didn't change it properly
19:09:16
stassats
how the hell do i, though?
19:17:39
stassats
got around bactraces using with-timeout
19:18:05
stassats
stream-error-position does read-char on stdin
19:19:40
stassats
(read-sequence "" *standard-input*) reads
19:30:39
stassats
maybe "" is printed with zeros as ""
19:39:44
|3b|
more windows test results, lots of stuff i don't recognize from last time i ran tests :/ https://gist.github.com/3b/714825e148ff605250cadac22fb57586
19:40:15
|3b|
though pathname stuff might just be intended changes without changing tests to match
19:49:29
stassats
setfilepointerex just hangs on stdin
19:51:18
stassats
on mac, (file-position sb-sys:*stdin*) => 2734487
19:51:31
stassats
that's suspiciously large
19:57:27
stassats
for windows, the fix is easy just return NIL on pipes
20:08:06
stassats
(cosh 90) decides at random what error to signal
20:09:45
stassats
ok, native-namestring got a bit broken
20:13:20
scymtym
and i tried to make things better, not worse
20:14:40
scymtym
some more successfully than others, it seems
20:15:15
stassats
can't break a few omelettes if you want to make eggs
20:16:28
|3b|
ACTION should save stderr too when running tests, commented gist with 'fatal error' from bit-vector.impure-cload.lisp
20:19:06
scymtym
stassats: what's the NATIVE-NAMESTRING damage?
20:20:53
scymtym
hm, i tested many variations of that specifically. must have missed something
20:21:55
stassats
just C:\\foo is enough
20:23:09
stassats
well, no need, i'm halfway through it
20:23:31
stassats
looks like two non-nested whens got into a cond
20:28:07
stassats
i think there's been enough fixes for windows (not just for regressions) that i might have enough extra motivation to post up a binary come release time
20:31:11
|3b|
ACTION wonders if the test hangs on windows got more deterministic at some point
20:32:49
|3b|
(or the non-deterministic hangs went away and were replaced by deterministic ones)
Wednesday, 22nd of November 2017, 20:34:41 UTC