Live data from Hacker News

iOS at Facebook [pdf]

static1.squarespace.com

71–80 of 111 posts

Re: iOS at Facebook [pdf]

#71

Earlier quoted context omitted.

Is it Toll-Free Bridging?

Yep, that was it. Man, I really misremembered that. It's an Objective-C thing, which makes sense, because it was for an iOS position. I knew the idea behind it, but I had never encountered the term. I do remember I ran into it again while preparing for interviews again a year ago. I'm deep in .NET territory for work and Python for fun right now, so my iOS knowledge has atrophied a bit.

Yep, I lost an interview about ten years ago in part because I had not encountered the magic term "duck typing" yet. Never mind that I had be making file-like and string-io objects in Python for years by that time.

Don't get me started, tech interviewers are so bad...

Re: iOS at Facebook [pdf]

#72
post #36
post #14

Earlier quoted context omitted.

The parent is not referring to those needs (which are legitimate). The presentation talks a lot about the hacking culture; how there is a lot of duplicate code, a lot of dead code, a lot of "temporary" code. We call these things technical debt. But it sounds like they just never fix them. Refactors never happen.

They don't call it debt. It is simply their way of keeping their agility with a huge number of developers. If the flip side is junk that mostly impacts install space, I don't see it as a bad trade-off. What alternative way would you suggest to keep your mega team productive and shipping features?

You can't avoid fixing tech debt. Eventually it impedes your ability to be productive. It happens, to some degree, to every software project. Someone else joked about "we wrote the app from the ground up" but that really does happen. You're not "shipping features" if you're rewriting the app.

Re: iOS at Facebook [pdf]

#74
post #2

So, FB app is bloated because they choose to be sloppy is what I read. Wonder how sustainable this is over the long term?

It's not sustainable. But since software developers have been dealt a good hand (demand-wise) in these decades, we get away with it. And call it "hacker culture", so it sounds hip and even correct.

Actually, I'd counter that it is sustainable -- for Facebook. They have so much user data and so much ad revenue that they can afford to throw engineers and money at the problem. Big companies can work hard and make things work, smaller shops have to work efficiently out of necessity.

Re: iOS at Facebook [pdf]

#75
post #2

So, FB app is bloated because they choose to be sloppy is what I read. Wonder how sustainable this is over the long term?

Product-wise, it's likely that in the long term they're going to break up the FB app into a federation of related apps. They already did that with Messenger.

Re: iOS at Facebook [pdf]

#76

Would be interested to see some comparison with the Android app which seems to have most of the same functionality at 1/4 the size with 25MB [1] compared to the 100MB of iOS. Seems like either iOS team at FB has significant cruft laying around or iOS doesn't scale as well as Android and FB is hacking around that. Possibly a bit of both but think it would be an interesting comparison in addition to the war stories abo…

iOS apps are shipped as fat binaries, so for one, part of that 100mb app is duplicated slices of binaries for armv7, arm64 and possibly, but not likely, armv7s.

Android apps enjoy bytecode that is getting JIT-ed on install.

Re: iOS at Facebook [pdf]

#79
Am I reading this right, there are interns and engineers running PHP in a debugger on production data? How else would he be able to memorialize his own account and trigger the email?

Re: iOS at Facebook [pdf]

#80
post #66
post #11

This didn't really tell me why Facebook needs 400 people to work on the iOS app and 75 commits per hour.

What's crazy is that WhatApp has 4 Android engineers and 1 billion downloads.

> What's crazy is that WhatApp has 4 Android engineers

Interesting. Any source for this claim? And also what's the full size of Whatsapp team?

Post reply on HN