github.com/moarvm/moarvm | IRC logs at colabti.org/irclogger/irclogger_logs/moarvm
Set by AlexDaniel on 12 June 2018.
nwc10 good *, #moarvm 07:31
MasterDuke partial success implementing pointer to current candidates and current guard tree last night. i traded a segv during the nqp build for an earlier segv during the nqp build 07:57
it just all depends on how one defines success...
nwc10 "it compiles - ship it!" is not good enough round here? :-) 08:00
timotimo it compiles ... but at what cost ... 08:57
travis-ci MoarVM build failed. JJ Merelo 'Merge branch 'master' of github.com/MoarVM/MoarVM' 09:01
travis-ci.com/JJ/MoarVM/builds/211005002 github.com/JJ/MoarVM/compare/e776d...d1665c5f3e
Geth MoarVM/solaris: 4 commits pushed by (Nicholas Clark)++ 12:01
MoarVM: nwc10++ created pull request #1412:
Get MoarVM building on Solaris		 12:03
nwc10 oooh, that's going to break nicely 12:04
I should have tested that branch *before* the PR 12:07
Geth MoarVM/solaris: 5e031c9175 | (Nicholas Clark)++ | 5 files
Add a probe for pow() corner cases with Inf and NaN.

Solaris doesn't default to being POSIX conformat, but NQP and the spectests assume that we are. It's not clear *how* to flip the defaults with gcc, so code around the problem.
Disable JITting `pow_n` if the C library pow() isn't conformant.
  (I don't have x86 Solaris to test, but given how thorough Sun's engineering
was, I assume that it's C library behaves identically to Sparc's, so likely it would JIT to the LTA implementation.)		 12:12
nwc10 that stands a fighting chance of working on non-LTA platforms 12:17
nwc10 good *, jnthn 16:25
jnthn o/
nwc10 \o 16:54
MasterDuke gist.github.com/MasterDuke17/a2f24...1e8d54455f i'm not sure if i'm missing something like a memory barrier or just haven't implemented this correctly? but i still get segvs very early in nqp build 17:02
lizmat and the last Rakudo Weekly News of 2020 hits the Net: rakudoweekly.blog/2020/12/28/2020-...evolution/ 18:56
nwc10 OK, welcome to crazy. 20:27
On Solaris, if you use /bin/sh to start a new process, it *will* update the environment variable PWD to the current directory
oh, unless the current directory is /tmp
(or I think, subdirectories of it)
no, seemingly it's set if cwd happens to be ~ or / 20:36
otherwise it's unchanged
leont How wonderful
nwc10 indeed
at least now after an hour of fiddlign with truss
I've figured out that my test C code was the problem 20:37
I need to exactly replicate the directory of the problem Rakudo spectest
well, most accurately it seems
either /
or a path that resolves to be the same as $HOME
because it's canonicalised to /export/home/nick 20:38
nwc10 this is ksh93, I think 20:45
patrickb I'd value some eyes on github.com/MoarVM/MoarVM/pull/1408 (and github.com/rakudo/rakudo/pull/4138 for that matter). 21:41
tellable6 2020-12-24T00:15:10Z #raku <tony-o> patrickb i'll merge winpass and that should get rid of the gcc requirement
2020-12-24T22:14:53Z #raku-dev <ugexe> patrickb: there are cpan examples in the Appveyor file
2020-12-28T19:00:09Z #raku <tony-o> patrickb can we remove the config stuff from the password branch and put that in a separate issue?
patrickb I love to be able to merge those PRs a good time away from the next release. 21:43

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!