Welcome to the main channel on the development of MoarVM, a virtual machine for NQP and Rakudo (moarvm.org). This channel is being logged for historical purposes.
Set by lizmat on 24 May 2021.
Geth MoarVM/trash_debugspam_for_azure: f958cb72c9 | (Timo Paulssen)++ | 2 files
turn backtrace for every gc run off again
 02:46
ugexe github.com/grpc/grpc/issues/37210#...2226809722 - i think this might be the issue i had with my appveyor builds failing 04:52
I wonder if we could just use 1.0 / 0.0 and -1.0 / 0.0 and drop the other variants for github.com/MoarVM/MoarVM/blob/570d...m.h#L9-L20 04:55
Geth MoarVM: ugexe++ created pull request #1838:
Improve NAN and INFINITY definitions for MSVC		 17:35
[Coke] ugexe: I tried to build that on my wintop, but the moarvm branch is old, and I had trouble doing the rebase. 17:45
... let me try one more thing... 17:46
(nuking my old moarvm windows install from orbit) 17:49
so if I start over, then try to do a git rebase origin/main in your branch, the merge goes sideways. 17:56
[Coke] will try to just apply this as a patch locally
[Coke] (this would be easier if I wasn't an idiot) 17:59
that last time, I tried to add your moarvm remote to *NQP* and then merge. wow. 18:02
finally got past all that and it dies trying to process the submodules in 3rdparty/cmp
ugexe can you post the full error? 18:11
it looks like its building in dev.azure.com/MoarVM/MoarVM/_build...8ddb1caf67 18:12
i wonder if we should switch `#define MVM_NUM_POSINF (DBL_MAX+DBL_MAX)` to `#define MVM_NUM_POSINF 1.0 / 0.0` on msvc like it is for other OS. i thought 1.0 / 0.0 was defined in C to mean infinity, whereas adding two DBL_MAX presumably just overflows (which to me seems like it wouldnt work right, but I'm not very knowledgeable about any of this either) 18:28
[Coke] sorry, that was on a different machine, I'll post the error I'm seeing this evening when I get back to it 20:13
Geth MoarVM/trash_debugspam_for_azure: 2ab93a89ed | (Timo Paulssen)++ | src/6model/reprs.c
debug output when die_no_attrs is called
 22:30
[Coke] ugexe: gist.github.com/coke/ddaa603d58a5c...330bfb7fb1 22:56
timo Found C:\raku\bin\moar.exe version 2024.07, which is too old. Wanted at least 2024.08-6-gac82e446f 23:02
you will need to update that, or other stuff will also just violently and randomly asplode, i sometimes find 23:03
Updating submodules .................................... fatal: transport 'file' not allowed
this is a security feature where you can't "git clone" from a relative file url or something like that 23:04
try `git submodule uninit` or what it's called, followed by `git submodule init` && `git submodule update`
may be required to `git submodule sync` to change the remote addresses for the submodules to the ones in the .gitmodules file that's checked in 23:05
[Coke] "need to update that" - I'm rebuilding it? 23:06
will try the submodule stuff
timo you seem to be rebuilding rakudo, not moar 23:07
oh it's --gen-moar 23:08
i keep forgetting that that exists
[Coke] timo: did deinit, init, update (failed) sync (ok) update (failed) - both failures are still "did not contain 3c5a..." 23:29
this appears to only happen on the ugexe/fix-msvc-infinity-nan branch - if I checkout moarvm origin's main branch, 'git submodule update' works fine. 23:31
... but if I switch *back*, the update works? weird.
at some point there's an entirely fresh copy of MoarVM in here as well, this isn't still the very old one I built my windows from last time 23:32
timo let me have a second look
* 3c5a72e - (HEAD) TRUE and FALSE don't exist everywhere (7 years ago) <Timo Paulssen> 23:33
[Coke] eh. It's building now that I did the update on the main/ branch
timo this is the commit that it's trying to get
[Coke] ok, new error: "cannot open include file: 'dynload.h' 23:34
I haven't built on windows in ages, so guessing this is also on main with my current setup (which also just had a forced upgrade to win11 classic)

Powered by Cro and the Raku® Programming Language.

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

Thank you!