From flussence@gmail.com

IRClogs for reference​: http://irclog.perlgeek.de/perl6/2014-12-29#i_9865603

I have this golfed-down example, tested on Rakudo 2014.12 JVM+MoarVM​:

  await Promise.anyof( start({ sleep 2*3; say 3 }),
Promise.in(2*1).then({ say 1 }) ); sleep 2; say 2; sleep 3;

I'd expect that to produce output roughly equivalent to writing​:

  for 1..3 { .say; sleep 2 }

But what happens instead is more like this​:

  sleep 6; say 3; say 1; sleep 2; say 2;

Other stuff I've found​: If I change the start() to a
Promise.in().then(), or swap the order of args to Promise.anyof, it'll
DWIM. Running the example with strace -ff generates 20+ megabytes of
output, which seems a bit fishy.

From @jnthn

On Mon Dec 29 13​:30​:31 2014, flussence@​gmail.com wrote​:

IRClogs for reference​: http://irclog.perlgeek.de/perl6/2014-12-29#i_9865603

I have this golfed-down example, tested on Rakudo 2014.12 JVM+MoarVM​:

await Promise\.anyof\( start\(\{ sleep 2\*3; say 3 \}\),

Promise.in(2*1).then({ say 1 }) ); sleep 2; say 2; sleep 3;

I'd expect that to produce output roughly equivalent to writing​:

for 1\.\.3 \{ \.say; sleep 2 \}

But what happens instead is more like this​:

sleep 6; say 3; say 1; sleep 2; say 2;

Other stuff I've found​: If I change the start() to a
Promise.in().then(), or swap the order of args to Promise.anyof, it'll
DWIM.

The scheduler was making some bad assumptions that Promise.in things wouldn't go doing things like sleeping or hogging resources for a while. While, of course, if you do that there will eventually be problems with thread exhaustion, there's no reason simple examples like this shouldn't work. So, fixed, and added a test in S17-promise/in.t.

The RT System itself - Status changed from 'new' to 'open'

@jnthn - Status changed from 'open' to 'resolved'