Fork me on GitHub
#cljs-dev
<
2017-03-24
>
anmonteiro00:03:49

@mfikes I’d be up for a box with that many cores 🙂 put me down

mfikes00:03:59

The problem with the Transit issue is that you essentially need such a box to repro the issue. Without it, you can't.

anmonteiro00:03:12

on a more serious note, are you sure that’s the commit which causes the regression?

anmonteiro00:03:24

it looks so.. inoffensive

mfikes00:03:33

Yes, I'm fairly certain. It almost, but not quite, deterministically fails every time with that commit, but not the one prior.

mfikes00:03:40

If anyone is interested in trying to repro downstream: 0. Get a dual-hexacore 1. git clone and cd planck 2. CLJS_COMMIT=86f26bf267f1c729313320c38e53e64450c0e68d script/build

mfikes00:03:48

Good call @anmonteiro — I finally got the previous commit to fail in the same way. So, I'm off to do more bisecting.

dnolen17:03:00

if you have any tickets with patches you want me to look at please let me know

dnolen17:03:43

@mfikes that issue looks pretty strange since Transit obviously knows how to handle symbols

mfikes17:03:15

Yes. It smells too much like the previous Transit issue that Chelimsky fixed.

dnolen17:03:25

are you sure this isn’t some weird classloading thing?

dnolen17:03:31

i.e. you have a target directory or something?

mfikes17:03:35

It is not a regression since it occurs with the previous release.

mfikes17:03:53

I run my clean-all script, but I

mfikes17:03:03

'll be sure to check that I'm not leaving behind any stale caches.