Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
  • Apr 14 21:55
    Datseris commented #166
  • Apr 14 20:20
    apbraun commented #166
  • Apr 14 20:18
    apbraun commented #166
  • Apr 14 20:09
    apbraun ready_for_review #166
  • Apr 14 19:57
    apbraun commented #166
  • Apr 14 19:56
    apbraun commented #166
  • Apr 14 19:55
    apbraun synchronize #166
  • Apr 14 19:55

    apbraun on fixed_mass

    tests for correlationsum_fixedm… (compare)

  • Apr 14 19:52
    Datseris commented #166
  • Apr 14 19:51
    apbraun synchronize #166
  • Apr 14 19:51

    apbraun on fixed_mass

    Some cleanup Change name and using statements Change to Neighborhood.jl and 1 more (compare)

  • Apr 14 18:48
    Datseris commented #166
  • Apr 14 17:38
    apbraun commented #166
  • Apr 14 17:38
    apbraun commented #166
  • Apr 14 15:00
    Datseris commented #166
  • Apr 14 14:54
    apbraun commented #166
  • Apr 14 14:48
    Datseris commented #166
  • Apr 14 14:35
    apbraun commented #166
  • Apr 14 14:06
    hkraemer commented #133
  • Apr 14 13:57
    xrisk closed #133
BridgingBot
@GitterIRCbot
[slack] <zenna> Thanks
[slack] <Tim DuBois> No, they travel. It's a true translation between one intersection and another one.
[slack] <Tim DuBois> agent.pos = (node, node, distance travelled)
[slack] <zenna> Ah..
[slack] <zenna> Ok, final question: do we know what cooridnate space map_coordinates is?
[slack] <Tim DuBois> One, sec, I'll get you the ref
[slack] <Tim DuBois> East/North essentially.
[slack] <Tim DuBois> This is all new to us, so keep the questions coming and open up issues for when things are not clear - we'll try to improve the docs.
[slack] <Tim DuBois> I've got to get to bed though.
[slack] <zenna> Ok thanks a lot
[slack] <zenna> great game btw!
BridgingBot
@GitterIRCbot

[slack] <Tim DuBois> Maybe this is a Distributed n00b question, but perhaps there's something happening with the @quickactivate macro in DrWatson.
Why does
@everywhere begin using DrWatson @quickactivate :ProjectName end
yield: ERROR: On worker 2:
LoadError: UndefVarError: @quickactivate not defined

But
@everywhere using DrWatson @everywhere @quickactivate :ProjectName
works fine?

[slack] <Datseris> no idea, i'm probably the biggest Distributed noob on this channel btw 😛
[slack] <Sukera> I suspect the @quickactivate macro does some eval shenaningans, which may be lifted outside of @everywhere begin ... end?
[slack] <Sukera> not a DrWatson expert though
[slack] <Datseris> Hm, not really, it is a really simple macro:
[slack] <Datseris> Oh, I guess the interpolation $ counts as eval?
[slack] <Tim DuBois> Ah, may just need an esc around things for hygiene.
[slack] <Datseris> no idea what esc does, feel free to PR!
[slack] <Tim DuBois> Low priority! I'll play with it later when I have the time.
[slack] <Tim DuBois> > feel free to PR
Will do.
BridgingBot
@GitterIRCbot
[slack] <Sukera> it's not that the interpolation counts is eval, it's that it'll be executed with the wrong qualification
[slack] <Tim DuBois> Indeed. A simple fix I think. Just need some time to test.
[slack] <Sukera> no wait, I'm not sure that is the real problem
[slack] <Sukera> it can't even find the @quickactivate macro itself
[slack] <Sukera> since using only works at the top scope, I suspect the using silently failed on the worker, and thus the macro wasn't available
[slack] <Tim DuBois> Yeah, could be.. Well, I have some investigating to do.
[slack] <Sukera> the @everywhere macro may just not see the using properly
BridgingBot
@GitterIRCbot
[slack] <jonas.isensee> Could this be because julia tries to expand the quickactivate macro before executing the first using statement?
BridgingBot
@GitterIRCbot
[slack] <Tim DuBois> Don't think this is an issue IN DrWatson, but something to keep an eye out for with Julia 1.6 (rc1 at least): collect_results works fine, but collect_results! hits a world issue when loading BSON data. There's an open issue on the BSON tracker that may be relevant: JuliaIO/BSON.jl#69 For clarity, this isn't a problem using v1.5.3.
BridgingBot
@GitterIRCbot
[slack] <Datseris> I personally have largely moved away from BSON.jl to JLD2.jl. Not only it happens that one of the devs is also a dev of DrWatson, but I also had several "ignored issues" with BSON in the past that made me lose trust.
[slack] <Tim DuBois> I essentially started using it since that was what's in the "Real World Examples" page. Good to know it's not the best choice. Have you had any problems with JLD2 recently? I haven't used it in a few years.
BridgingBot
@GitterIRCbot
[slack] <Datseris> Oh good point, we should be updating some of these examples to use JLD2.jl if possible. I think the only limitations are with respect to what you can save? I (so far) only save numeric data so I don't have any problem. Perhaps there are some limitations with saving things like closures. Maybe @jonas.isensee can point you to a document that compares the two
BridgingBot
@GitterIRCbot

[slack] <jonas.isensee> Yeah, I've adopted JLD2 and am taking care of it now.
JLD2 puts a lot of effort into producing portable / robust files.
Currently it is not possible to store closures / anonymous functions / method instances.

( It will store functions in the sense that it can store e.g. [sin, cos, sum] but it won't serialize any methods attached to it)
I'm not saying that it can't be implemented, but it is definitely hard.
(E.g. how do you serialize a method when you don't have access to its code? Or how would you ensure that this could run on a different machine?)

[slack] <jonas.isensee> AFAICT there is no way to create portable serializations of method instances anywhere in the julia ecosystem at this time. (portable as in runnable in different julia versions or on a completely different machine)
In those cases, when you really need that, Base.Serialization may be the best option at this time (?) (which again is not portable)
BridgingBot
@GitterIRCbot
[slack] <Tim DuBois> Cool, thanks. I certainly don't have any need for these things at this time - really only storing data. I'll switch up my implementation!

[slack] <jonas.isensee> I am not aware of any such comparison.

As far as I can tell, a lot of the serialization packages (e.g. JLD2, JLD, BSON)
had a lack maintenance for too long..

[slack] <jonas.isensee> Feel free to open an issue if you run into any trouble.
[slack] <Datseris> very true, i was really mad at the state of saving data with Julia during the last year of my phd
BridgingBot
@GitterIRCbot
[slack] <Tim DuBois> Just some more info here. I thought it was a v1.6rc1 issue. But turns out BSON.jl upgraded to 0.3.0 recently and loading any of my bson files fail in v1.5.3 is well. If I revert back to BSON 0.2.6, things are fine.
BridgingBot
@GitterIRCbot
[slack] <Datseris> Damn it, @quickactivate does not work with Jupyter notebooks
[slack] <Datseris> come oooooon why
[slack] <Datseris> not even cd(@__DIR__) works with them.
[slack] <Datseris> ah lol: JuliaLang/IJulia.jl#915 time for recurring nightmares I guess
BridgingBot
@GitterIRCbot
[slack] <Datseris> Agents.jl reached 200 stars :party_wizard:
BridgingBot
@GitterIRCbot
[slack] <atiyo ghosh> well-deserved too. thank you maintainers!
BridgingBot
@GitterIRCbot
[slack] <Tim DuBois> Agents 4.2 has just been released with some big additions. Thanks to @Aayush Sabharwal, we now have a pathfinding module! Reworking of parallel model runs, distributed parameter scans et al now give users a full suite of ensemble tooling.