<SeanTAllen>
unless its related to signals specifically
<SeanTAllen>
in the PR i linked to earlier are instructions for using pthreads instead of signals for "sleeping" shceduler threads
<winksaville>
So want me to retest and abandon my current debug session?
<SeanTAllen>
¯\_(ツ)_/¯
<SeanTAllen>
seems like we might have a hanging issue
<SeanTAllen>
and we have the process manager bug
<SeanTAllen>
pick whichever you want to chase after
<SeanTAllen>
fixing either would be good
<SeanTAllen>
process manager might be easier to track down
<winksaville>
So I'm doing "both" at the moment, I was really saying abandon the gdb session. i.e. ctrl-c out of stdlib loop and recompile and retest
<winksaville>
Other info: ponyc was compiled on my Arch Linux system with llvm 5.0 and default_pic=true. stdlib was compiled with -d and running --sequential
<SeanTAllen>
ah well
<SeanTAllen>
llvm 5.0 is also experimentally
<SeanTAllen>
i assume that would have no impact here
<winksaville>
This is on my desktop system
<winksaville>
which has 12 HThreads on my laptop with 4 HThreads it hasn't hung after hours of running.
<winksaville>
Anyway, I'll keep trudging along and see what I learn.
<SeanTAllen>
huh
<SeanTAllen>
interesting
<winksaville>
Anyway, I'll compile for pthreads next and see what happens
<SeanTAllen>
cool
<SeanTAllen>
thanks for taking on the process manager issue
<SeanTAllen>
definitely a big win once we fix that
<winksaville>
np, its a way to learn something new, I assume you're on the east coast of US, so I'll let you go.
<SeanTAllen>
i can help you through your problem though
<SeanTAllen>
"receiver type: Eapproximator tag" <-- this means e_approx is a tag. because its an actor. you can only send messages to actors. aka call behaviors on them in an async fashion.
<SeanTAllen>
"target type: Eapproximator box" <-- you are trying to call a synchronous function. not allowed.
<SeanTAllen>
In this case I think you want your Eapproximator to be a class. but its hard to tell based on your example.
<SeanTAllen>
Are you just learning or is this part of a larger program?
<koczurekk>
Oh, I get it now! I've read about actors in official tutorial but it only said that behaviors are allowed and I didn't find anything about normal functions being banned.
<koczurekk>
I'm still learning.
<SeanTAllen>
so the important part is
<SeanTAllen>
actors, you have a tag reference to them
<SeanTAllen>
thats opaque
<SeanTAllen>
you can use it to compare identity
<SeanTAllen>
and send messages
<SeanTAllen>
nothing else
<SeanTAllen>
also, welcome!
<koczurekk>
Okay, thanks a lot. :)
aturley has joined #ponylang
koczurekk has quit [Ping timeout: 260 seconds]
aturley has quit [Quit: aturley]
SenasOzys has quit [Ping timeout: 268 seconds]
SenasOzys has joined #ponylang
jemc has joined #ponylang
SenasOzys has quit [Ping timeout: 248 seconds]
koczurekk has joined #ponylang
SenasOzys has joined #ponylang
jemc has quit [Ping timeout: 248 seconds]
jemc has joined #ponylang
aturley has joined #ponylang
koczurekk has quit [Ping timeout: 260 seconds]
aturley has quit [Quit: aturley]
jemc has quit [Ping timeout: 260 seconds]
jemc has joined #ponylang
koczurekk has joined #ponylang
<winksaville>
Why is stdlib always --sequential ?
koczurekk has quit [Ping timeout: 264 seconds]
<SeanTAllen>
winksaville: because there are tests that fail on Travis if it isnt run as --sequential due to resource constraints and timeouts
jemc has quit [Ping timeout: 240 seconds]
<winksaville>
Would it be more stressfull to run it without --sequential here at home?
<SeanTAllen>
not sure what you mean by stressful
koczurekk has joined #ponylang
<SeanTAllen>
for finding the process monitor issue i cant comment on if it is more or less likely to make it happen
<SeanTAllen>
all process monitor tests are run in an exclusion group
<SeanTAllen>
which means even without --sequential
<SeanTAllen>
they will only run 1 at a time
<winksaville>
I'll give it a try at some point, I assume we'd expect it to work with or without --sequentail, correct?
<winksaville>
(modulo resource constraints)
<SeanTAllen>
yes
<SeanTAllen>
basically
<SeanTAllen>
when you run without sequential
<SeanTAllen>
it tries to run all the tests at once
<SeanTAllen>
and with constraints resources
<SeanTAllen>
some timeouts are more likely to be hit
<SeanTAllen>
i plan on fixing that
<SeanTAllen>
but jemc and i have a disagreement on it so i put it aside for now
koczurekk has quit [Remote host closed the connection]
samuell has quit [Quit: Leaving]
rurban has joined #ponylang
rurban has left #ponylang [#ponylang]
mson has joined #ponylang
khan has quit [Quit: khan]
SenasOzys_ has joined #ponylang
SenasOzys has quit [Read error: Connection reset by peer]
lolman has joined #ponylang
<lolman>
is there a way I can get a complete list of functions that have special builtin behaviour? like create, apply, add, eq, has_next, serialise...
<SeanTAllen>
What do you mean by special behavior? Lolman
<lolman>
well the compiler has syntax sugar based on a create function
<lolman>
and apply
<lolman>
and the rest of them
<SeanTAllen>
sugar.c probably has most of what you want lolman
<lolman>
Thanks!
<SeanTAllen>
you're welcome
dipin has quit [Quit: dipin]
SenasOzys_ has quit [Ping timeout: 240 seconds]
mson has quit [Quit: Connection closed for inactivity]