Future directions of Julia

Dream big Chris!

3 Likes

Maybe there are people in the community who prefer Julia not to be adopted by industry and feel it should only be used for research purposes. I don’t know - just figured I’d put that parenthetical statement there to be political.

I don’t know if I’m joking or serious either, but it’s good to have goals.

4 Likes

Is there an issue tracking these (maybe with v2.0 tag)?

There’s a few related ones:

https://github.com/JuliaLang/julia/issues/35899
https://github.com/JuliaLang/julia/issues/30945

2 Likes

Can you explain what that is?

He created the following PR to ArrayInterface where he explains and implements some of those ideas:
https://github.com/SciML/ArrayInterface.jl/pull/70

1 Like

13 posts were split to a new topic: Towards improving community diversity

Let’s keep this thread focused on future directions and changes to the language itself.

11 Likes

That’s definitely better one.

Every few years I approach my most missed, most hoped-for, broadly coopting want for Julia: That the strengths now well painted within our language be applied elegantly to realize GUIs just work. To be constructed with ease, cooperatively and collaboratively, performant with aplomb and beauty.

8 Likes

Together with creating an easy-put-together, easy-to-install, stand-alone application.

1 Like

To be fair, is there a language which does this? I find nontrivial GUI programming tedious in all languages I know.

Not saying that Julia cannot do better, just that it’s something which will require quite a bit of work.

8 Likes

To be fair, is there a language which does this?

Yes, I would say MS Visual Studio with C# is that.

1 Like

Well, if you drop performant tcl/tk and pygtk are actually frameworks on ‘slow’ interpreted languages which can do a lot of non-trival stuff more or less out-of-the-box - cross platform.

But i was about to write a similar commen, GUI is never easy. I’m doing GUI (it’s not my major) since Intution/AmigaOS with some emphasis on X11/Xaw and GTK in mix of languages and yes, you usually need to learn some concept and apply them in the respective language.

I sometime think, that julia with multiple-dispatch could make some easier API, as you just call an operation like “pack” and you don’t need to specialize to Button in Box in Grid in Layout in Window in Screen.

But in general this belongs to infrastructure and the incentives to solve this for julia are low.

3 Likes

To be fair, is there a language which does this?

Yes, Self, Squeak, Snap! each provide this with Morphic .

1 Like

Limited to Windows, ain’t it?

Don’t have self experience but I heard that Mono+C#+.NET does a good job on linux.

1 Like

This is the most painful part for me. No matter how much we perceive the benefits of the language in our work and research, most people are just in a mindset of never switching out from Python. In the case of Matlab it is super easy to convince people that Julia is the way to go given the associated costs.

If people can’t feel it in their pockets, they won’t even try, unfortunately. I think this is correlated to the fact that the Julia community started as a group of curious PhDs willing to explore. Most people aren’t in this mindset and I couldn’t figure out a strategy that is effective in my own lab yet.

Are you aware of any company in which people get bonus to try explore new languages? I could borrow some ideas…

3 Likes

I can’t get people to switch from matlab. The Simulink and other packages absolutely dominate all other competitor offerings. But even some of the people not using simulink, say, doing data science, won’t switch because they know their packages and they often don’t know either the science or the programming - so they’d be in real trouble switching… Most bosses I’ve had see a 1k expense for a tool to be throw away money, it’s only once they start to approach 10k that they start needing serious reasons for it. Small tech. starts ups running single screen, nucs, ubuntu, etc though feel the costs heavily. But they also tend to go pure python(interview experience only).

I have not worked for any company that encouraged people to explore new languages. Jeese it’s hard to even negotiate time for personal/skill development (failed to do that as well)… Even back in grad school: some groups used Matlab, some used R, and never the two shall mix. I would be interested to hear about companies doing that though.