Ruby Conference

guild reminds me of maplestory

>>56488977
>It's going to lay out what Ruby 3.0 looks like, including how they plan to make it 3x faster
Based on what I know about Ruby, I don't really see how this is possible.

>>56491044

JIT. That'll give it a reasonable speed boost.

>>56491385
>jit

>https://blog.heroku.com/where_will_ruby_go_now_talking_with_tenderlove_at_railsconf
>Yeah, exactly. You can flip that with flags and stuff. So I don't really see how memory usage would be that much of an argument. Though, when you look at the stuff that Chris Seton has done on Truffle Ruby, so Truffle + JRuby — that thing is super fast, but you'll notice that in presentations that they never talk about memory consumption. They never talk about it.

I don't think ruby jit with low memory consumption is possible at this point. Unless they can make Mike Pall working for it, some how.

>>56488977

This whole 3x3 meme is just too late.

I really like Ruby as a language. I'd much rather write Ruby than JavaScript.

But a three times performance increase hailed in a major new version will still put it long behind many other platforms.

They should at least be targeting 10x if they want Go to stop stealing their users.

>>56491523
even rpg maker has stopped using ruby for javascript.

well, you still can write prototype in ruby then change it to crystal later for performance though.


If only Crystal has support for windows.

>>56491044
this has already been discussed in detail: https://www.youtube.com/watch?v=pE4LVSESyXc

>>56491385
They talked about that back in the 1.8 days, but as I recall, other structural improvements were deemed easier to implement, with more speed benefit. The 1.8->1.9 speed jump was amazing. I don't know if they could pull off the same trick with 3 again.

I am only interested in this conference if the organisers have assured racial, sexual and religious diversity

>>56488977
Are eggplants and dongles banned?

>>56492071
That's about the most obvious bait one ever seen.

>>56492143
I don't think it's bait, just irony

>>56492071

It's in Japan. Japan doesn't really do "diverse."

>>56491523
The whole problem with Ruby is that it's a really, really comfy syntax married to an awfully complex structure underneath. How can you get performance out of a language with almost no real basic data types, or one that has to check variable names for context? I'm not a language expert but Ruby's speed seems limited by design to me. You can't have luxuries baked in without the performance hit.

Who the fuck uses Ruby anyway. Is this 2011?

>>56492241
ever heard of Truffle?

>>56492241

>almost no real basic data types
Fixnum, Bignum, Float, Array, String, Hash... Ruby has plenty of data types. What it doesn't have is primitives. That said, some performance improvements have been made despite the system being wonky. For instance, the interpreter can tell an object is a Fixnum just from its object ID, and can use that object ID to extract the integer it represents without having to further access memory.

>one that has to check variable names for context
You do scanning/parsing once, and it's typically not the most computationally expensive thing to do. Also, Ruby's name rules only require checking the first and second character (first for everything other than class variables) to determine everything you need to know.

You would be right in saying that some of the design features of Ruby will make optimization difficult, but some improvements can still be made. Recall that Javascript is also relatively dynamic, but has approached many other JIT-compiled languages in performance. If Ruby had a native JIT compiler, it might use more memory, but it also might have faster execution speed.

>>56492606
Believe it or not, there is more to Ruby than just Rails.

>>56493182
>the interpreter can tell an object is a Fixnum just from its object ID
2*x + 1 right?

>>56493298

Yep. On 64-bit Ruby, there's also some system for certain floating point numbers, but I'm not sure what it was.

>>56491523
Ruby is already beating Python at most benchmarks. If you triple its speed it won't make it C-level fast, but it will make it much faster.

The main thing 3.0 can bring is true parallelism. Once that happens, Ruby will be significantly better at Web-related tasks, which is awesome.

More importantly, continuing to increase speed shows that the Ruby team is growing in competence, and that Ruby is still vibrant as a language. It's great for marketing, at the very least.

>>56488977
this multi threading proposition actually makes sense, and is pretty decent, i could get behind this.

>>56488977

>>>56492606
>Believe it or not, there is more to Ruby than just Rails.

Been using ruby for 10 years. Have never touched rails. Can concur.

>>56488977
>It's going to lay out what Ruby 3.0 looks like, including how they plan to make it 3x faster, and possibly add some sort of type system.
So it's basically Crystal, except probably less thought through?

>>56491467
>Unless they can make Mike Pall working for it, some how.
He's good, but remember Lua is a much smaller language than even JS, which means, it would at least take 10 Mikes and 5 years to get to a usable JIT compiler for Ruby or Python.

>>56491467
Ruby is hard to JIT, but I think that trading memory consumption for speed is okay in some cases.

File: 201940129408214.jpg (99KB, 620x349px) Image search: [Google]

99KB, 620x349px

>>56488977
I would have gone but it had a "code of conduct".