Facebook's HHVM To Focus More On Hack, No Longer Focusing On PHP7 Compatibility

[phoronix]

Phoronix: Facebook's HHVM To Focus More On Hack, No Longer Focusing On PHP7 Compatibility

Some interesting remarks today by Facebook's HHVM/Hack language team as they plot their future agenda...

http://www.phoronix.com/scan.php?page=news_item&px=HHVM-PHP7-Focus

[Steffo]

Didn't the latest benchmark show, that PHP 7 is more performant than HHVM? Why not invest in PHP 7?

[LEW21]

Google, Apple and Microsoft each have their own languages, so Facebook has to have its own too!

Except that the others have engineers capable of creating good languages, and Facebook does not.

[mrtruxton]

I had some experience with HHVM. The main reason to work on this was a lot of developer hype. As a sysadmin i tasked myself with facilitating the devs who wanted to give it a spin. The main platforms we piloted where Magento and Wordpress. Both platform kind of worked as long as you didn't add too many plugins, but they all had strange quirks. Date functionality would not work correctly, handling of certain pieces of code would yield different results. Unit tests failed, Etc. From our initial testing it became obvious that this was not something we could simple use as a performance update to our customers. Instead we could only consider it for new customers where we would have to double down on compatibility testing. A high-end customer with large scalability requirements and a big budget. In the end it the announcement of PHP 7 made us hesitant to act and in the end we decided to abandon the plan all together.

Two things still speak in favor of HHVM however. Hack and memory consumption. The strong typing features of hack can help you make some really solid applications from a data processing standpoint. Also the memory consumption of hack is a lot lower and a lot less spiky.

Still, in the end, even with the remaining benefits, we will probably never look at HHVM ever again. Unless a great open source application gets build in it. ( and does not run on PHP7. ) PHP7 has also started a trend of implementing strong typing features as well. If this continues to improve it might remove even more than ever the need to switch to HHVM.

In the end i am still sad to see HHVM go down this road, but maybe it is for the best. They where a great catalyst for the PHP community to get of their asses and improve some longstanding performance issues. I can only hope the PHP community will not slack down now that their main competitor is going to disappear.

[Danielsan]

Why not invest in PHP 7?

Because FB decided to play dirty with the open source, thus because its games has been revealed now it decided to move straightforward in that direction, let's see how many devs are going to follow FB in this no outlet path...

[birdie]

Because FB decided to play dirty with the open source, thus because its games has been revealed now it decided to move straightforward in that direction, let's see how many devs are going to follow FB in this no outlet path...

Or maybe just maybe they wanted the features or/and the performance which PHP5 couldn't provide and PHP developers couldn't accept their patches. It's always "bad" companies and good open source. Lovely. I mean the amount of blind hatred and fanaticism is just staggering.

[boxie]

Or maybe just maybe they wanted the features or/and the performance which PHP5 couldn't provide and PHP developers couldn't accept their patches. It's always "bad" companies and good open source. Lovely. I mean the amount of blind hatred and fanaticism is just staggering.

you gotta love clean and pure idealism, it works great in theory - but the real world is a rather messy place!

good luck to Facebook and their Hack language. If they can build it up and add compelling features and convince developers/frameworks to switch over - brilliant - if not, then they will have their own open source language that they use.

[nazar-pc]

Well, I think it is time to finally drop HHVM support in my projects

[microcode]

To be honest, with their frontend developers (seemingly) slowly moving toward Reason (basically OCaml), I wonder why they don't just ditch Hack altogether and go toward OCaml/Reason on the backend as well.

[Danielsan]

Or maybe just maybe they wanted the features or/and the performance which PHP5 couldn't provide and PHP developers couldn't accept their patches. It's always "bad" companies and good open source. Lovely. I mean the amount of blind hatred and fanaticism is just staggering.

They are free to to do whatever they want but their license is pretty iniquitous, they can have even the best 'pros' on the market but I believe no one is going to follow them.