3 ms·
I see this as being useful if you already have these two things: 1) A large PHP codebase and 2) A CPU-bound app with scaling issues If you're starting a new
by mrshoe 17y ago
I see this as being useful if you already have these two things:
1) A large PHP codebase and
2) A CPU-bound app with scaling issues
If you're starting a new app, I don't see why you would choose PHP. And I think most web apps are database-bound rather than CPU-bound. HipHop seems incredibly useful for Facebook, but are other people here on HN excited about using it?
- lanstein 17y agoI have a CLI application that's IO-bound. I'm extremely excited. Furthermore, I'm sure this is exciting to anyone who has deployed Memcache, and now has an application that is CPU-bound.
- ighost 17y agoI wish memcache changed my IO bound app to CPU bound.
- ivankirigin 17y agoYou can optimize on multiple fronts. Just because facebook needs to scale data access doesn't mean it shouldn't scale processing time too. As the post mentioned, this has had a huge impact. [i work at facebook, comments on this site are my own opinion, and I'm not intimately involved in the project] But generally I agree about new projects. If you have a favorite language, use it.
- coliveira 17y agoThere is no need to switch to PHP only for this reason, but there is a huge number of apps already using PHP. It will be a great benefit to them. Also, it will become another factor for people considering developing a new application: if a language can quickly be converted to C++, this is easier than writing extensions in C/C++ directly.
- mikeryan 17y agoI think this is where many run into issues with this story. This project was never really meant to be something to help the PHP community as a whole. It was meant to be a tool to help Facebook and to that end they've succeeded very well. 3 engineers + 2 years = 50% CPU improvement without having to rewrite the whole PHP codebase from scratch? That's a great achievement for Facebook but YMMV.
- staunch 17y agoSeriously. $100k * 3 devs * 2 years = $600k to buy half as many app servers in the next year(s)? They saved Facebook tens of millions of dollars. I should imagine their next performance evaluation will be quite smooth. This is what people mean when they talk about great hackers being 10x as productive...
- robryan 17y agoIt would have been a great freedom for the devs, at a company of facebook's size it wouldn't have been disastrous had they not really achieved a great enough performance boost to justify it's use.