This is really interesting, I've seen this pattern somewhere. IMHO, CoffeeScript had a really similar effect on the development of JavaScript / EcmaScript, but then, it quickly started going extinct. Same with IO.js vs. Node.JS, which was eventually re-integrated into the main line.

It's interesting how apparently sometimes you just need to step aside to break old conventions and demonstrate new capabilities, but eventually, you'll need the support of the community to pull it through: Plus, it's always a gamble. Python barely made it over to v3 after 10 years. I'm not sure if Perl will ever cross the gap or if Perl 6 fragmented the community for good.

Just some random thoughts though on the dynamics of open source, I'm glad for the experience of PHP at the beginning of my career, but I wouldn't touch it or any derivatives any time again having free choice. Too much Heinz™ design: Grown, not made.

These fork-like efforts are able to take the risk of the experimentation without putting the mainline project at risk, and their successes allow the supporters within the original project to point to it as a viable path and gain the needed mindshare to pursue something achieving the same goals.

It's a similar model to larger corporations allowing innovation to occur in startups and then buying them out once the model or technology is proven to a degree.

I think CoffeeScript made transpilers mainstream (though it wasn't the first) and i appreciate that (having a sane language like clojure in your browser is amazing), but damn that syntax was confusing more often than not. Sometimes less is not more.

My thoughts on this from practical experience at Perl conferences and Open Source conferences in general: many (former) Perl 5 programmers are looking forward to finding a use-case to use Perl 6 for a project in their environment. And some have done already, and put the result in production, which may be specifically easy with micro-services using Cro (https://cro.services).

Of course, there are (former) Perl 5 programmers who still think Perl 6 is the sole reason for Perl's demise. These are fortunately very few, but alas also very vocal.

It really boils down to the implementation. Also, you need community support. What really helped HHVM mature is that by using PHP for something large scale, they were able to see its shortcomings. Most other forks are not driven by that. As such, their upsides are questionable, or they just offer some performance boosts, that are achievable even without the trouble of forking.

Speaking of fragmentation, we still have 3/4 of users on PHP 5.x, and support for v5.6 (the last 5.x release) is coming to an end soon!

And honestly the jump from php 5.6 -> 7 is pretty painless. Its probably the lack of tests for a lot of old 5.6 software that gives people fear of upgrading. php development has gotten a lot better.

I think support for 7.0 is ending soon too.

Yes. It reaches the end of its support at the end of this year, and PHP 7.1 next year.

Rasmus Lerdorf actually acknowledged HHVM's role in numerous occasions, especially with reference to the development of PHP7 (Edit: sorry I tried to find a link but no luck so far).

Well, Hack has a lot of advanced features that is missing from PHP7. Proper data structures (vector, map, set, etc...) including immutable ones, generics and variance, async/awaitable, etc...

I would love love love for all these to be in PHP.

We'll probably get them eventually, it looks like the typed properties RFC is going to pass this time around.

Which is great for me :).

Yeah, at a previous job we used Hack around the time PHP7 was released and switched over to PHP7 pretty quickly.

Performance and type checking were the only benefits over PHP5 and with those (mostly) gone, you were just left with all the downsides like all the good tooling for PHP being incompatible with Hack and, and general compiler/interpreter errors that were never getting fixes.

I mean, I think people are somewhat missing the fact that it's not like these worlds are separate or competing. Some people have worked on both PHP and HHVM, and it seems like they've always acknowledged that they have different goals. From the comments in the post of the Hack announcement (before PHP7):

Q: Do you imagine a future where Hack will merge back into PHP (like PHP 7), in the same style that Beryl & Compiz then rejoined? Or does the team intend for the two to always be adjacent-yet-separate?

A: HHVM developer & PHP runtime developer here. I've got hands in both runtimes and all even I can say is: Maybe. I think the most likely outcome is that PHP will adopt some of HHVM's additional features, but remain a separate project. IMO that's a great outcome, since we'll both likely drive the other to be better.

EDIT: Source, https://news.ycombinator.com/item?id=7436557

HHVM can be more experimental than PHP so there's still potential that new features in HHVM get considered and merged into PHP, kind of like Fedora/Redhat or Neovim/Vim

Very true, especially as type hinting for properties looks to be coming in PHP 7.4!

https://wiki.php.net/rfc/typed_properties_v2