Brief History of JIT in MoarVM for Perl 6

> It is a really nice language based off of all the tutorials and presentations I've watched. It just needs some better performance before I'll really start using it

I'm also very impressed in how the language is turning out. It's a pleasure to write and read. I know most people really don't like Unicode operators, but I find them to be very useful. They're short and concise, and allow those writing mathematical equations to stick to their regular symbols. And for those that don't know how to do Unicode on their OS/editor, there's ASCII equivalents.

Performance-wise, it seems to be fast enough for all my personal projects, and it's getting faster every release. Is there a certain piece of code that you want to have running within a certain amount of time? That would allow you (and Perl 6 devs) to have a goal to look at, and work towards.

Many years ago, it was. Parrot is a separate team, and was designed to be the "mother of all VMs" with Perl 6 as its sort of halo language implementation. It split off as the Perl 6 project matured and needed to own the vm/compiler.

There is an intermediate language the Rakudo compiler is implemented in called NQP. That split from the Parrot project to support several VMs as the stuff emitted as the "compiled" code. For a while Parrot and JVM were supported simultaneously. With the addition of MoarVM the project moved away from supporting Parrot. There are other targets in development including JavaScript and most recently Truffle/GraalVM https://github.com/perl6/nqp/commits/truffle Essentially a lot happens within the Perl 6 project at a pace that's kind of surprising if you keep an eye on it.

Since H. Merijn Brandt started working on the Perl 6 version of Text::CSV, he has kept a log of how long a specific test (referred to as test-t) too to run.

https://tux.nl/Talks/CSV6/speed.log

Since then, Rakudo Perl 6 has become more than 100x as fast, at least for this test. And more than 250x as fast using a concurrent version of the test (which was as simple as putting a `.race` in the code).

This makes it still slower than the Perl 5 version, which uses XS (all parsing logic written in C). But the gap is closing with another round of optimizations to be merged later this week after an extensive refactor that took about 2 months. Of which you will most certainly see a report on https://6guts.wordpress.com , the blog of Jonathan Worthington.

MoarVM has recently gotten the ability to write optimization plugins in higher level code. This should make it easier to create certain optimizations as they will have access to the syntax tree.

So it may be that the rate at which it gets faster has gotten faster.

Quoting the blog post this HN article is supposed to be about[1]:

PS: You might read this and be reasonably surprised that Rakudo Perl 6 is not, after all this, very fast yet. I have a - not entirely serious - explanation for that:

All problems in computer science can be solved with a layer of indirection. Many layers of indirection make programs slow. Perl 6 solves many computer science problems for you ;-)

In the future, we'll continue to solve those problems, just faster.

[1] http://brrt-to-the-future.blogspot.com/2018/07/perl-6-on-moa...

It really depends. The question really should be, "Is Perl 6 fast enough?" For my purposes, the answer is almost always yes. I can imagine some applications where the answer would be no, but then I would probably be looking at C/C++ for a solution, anyway. Language bench-racing has really begun to look counter-productive to me.