The only thing I can think of that I use a factory for is database connection pooling. It works exactly the same way that making a connection directly would, except that if there already is one, it gives me that one, and if I'm over the limit, it makes me wait.

In python you actually can do this without a factory, but using a factory is simpler.

Oh wait, I also use it in one spot to delay instantiation of an object to the first time you use it. Basically I'm just delaying the import of a 3rd party library because it has side effects and I want them to occur after, not during, startup.

dont even get me started on factory of factories

suzzer,

yea. a lot of times it feels like an excuse to me to create bloated code all in the name of "abstraction"

Using SourceForge is bad right? I know they are super scummy, and I have some momentum on my side to make the change.

I just don't know if for some reason people still actually use them for a reason that I don't currently understand.

Most factory code just looks like oop spaghetti to me

the deeper i get into CS, the more i hate OOP. i really liked it when I was a new programmer.

now I'm programming some pretty crazy stuff in C using just structs and I think "who tf needs elaborate object classes anyway"

Good video on the subject if you haven't seen it.

Casey's rants on OOP are fun

OOP stuff has it's place. In a lot of cases, a project gets jammed into an OOP mold when it's not really needed, though. That is, even when using an OOP language you can just use classes as buckets of data and functions without going nuts about inheritance and all that

Actually in a lot of cases I am good with a language that has the notion of "interfaces" over inheritance, which is probably approximately what the video about composition over inheritance about. "interfaces" in the java class term lets you say "the class has this list of functions that take these arguments" and that gets you 90+% of what I want out of OOP.

Feel like this covered it pretty well.

OOP with inheritance is a disaster.

And almost surely everything you learned in school was bad OOP, and you were doing procedural programming through a facade of objects. And if you're doing that, of course it's bad. You're just adding a needless abstraction and lying to yourself, rather than doing the thing you are doing -- procedural programming. I see Casey coming from that POV.

That said, I think it's possible to do OOP well, but it's difficult, and I think you need to understand functional programming first, and if you do that, you won't need anyway...

Like all things in the CS field, it's horses for courses. The low level, embedded controller software I used to write on 68HC11 micro-controllers was not an environment where OOP would have been useful.

I can see how OOP is more applicable, however, when I'm coding in Swift on iOS.

It probably helps when the language is built with everything being objects instead of the hodge-podge of bolting it onto existing languages.

Agreed it helps, but it's not sufficient. Everything is objects in ruby, and the majority of ruby code is still bad OO or procedural dressed as OO.

i feel like the latest snapchat update will eventually be a case study in terrible UI design and misunderstanding your user base.

I use Snapchat for work, won't go into why. It's a candidate for the worst app I have ever used. No idea why anyone uses it by choice.

it was never great - but they took the one great feature they had, removed it, and somehow made the UI even worse than it was before.

it just reconfirms a long held belief i've had, is that application developers should be avid users of the app they are developing. there's no way any one in charge of this update used it regularly and was like "yea, this is way better than before."

Yeah the changed version is worse, it was always terrible though. Incredible how much unintuitive behaviour they managed to pack into a simple app.

what gets me is like where you press on the new "story' all in one screen is supposed to affect whether you can send a snap back or view the story but even that doesn't work how it's supposed to, like i still can't figure out even with the tooltips what the hell I'm supposed to do to send a chat or view someone's story.

I cant even tell if the severely reduced amount of stories I’m seeing is a feature or just because people aren’t using the app since the update. I know I’ve only really opened it like five times since they updated it and i used to obsessively check it when i was bored and had nothing else to do.

I finally figured out the stories a day or two ago. You go to where you used to only get actual direct snaps, that's where the people's stories live now except for a couple that get put where all the stories used to be.

You might have to click on the circle avatar at the left of the cell to view the story, because if you click the main body of the cell it just opens the box for you to send THEM a snap.

I've always thought that part of the appeal of snapchat for teens is figuring out how to use it, swapping tips, and being in the club once you figure it out.

You guys don't think they AB tested this change? They almost certainly did, and it probably won in whatever metric they care about. (probably revenue)

There's a lot of momentum behind picking the new thing after you spend a $million developing it and some exec's butt is on the line. I've never seen AB testing result in "nope, let's kill it" for any major redesign. Maybe small things, or maybe tweak the redesign a bit.

Maybe snapchat is different but I kinda doubt it.

^100% that

The biggest problem I have with snapchat's app is that on Android it is really slow. They've got so much packed in there, I just don't think a phone can handle it. From tapping the app icon to taking a photo is ~5-10s. That's just not fast enough when you're trying to capture spontaneous moments a lot of times.

From my experiences doing website redesigns for tech companies focused on lead-gen/conversion optimization, when you do a full scale redesign you end up at the whim of exec opinions to a startling degree. Since its physically impractical to comprehensively test every new part of the experience (and much of it isn't clearly testable because so much is changing) you end up with explanations and stories for why the new design will be better. While a few critical metrics are tested and the larger end-to-end experience is neglected.

The best redesigns are iterative where you can isolate changes and compare them on a much much larger number of factors. I don't use Snapchat, but I've heard there is a new change that "celebrities" are available by swiping left and your friends by swiping right. If that was a single change, it may impact 50-100 metrics easily, and take a while with a significant sample size to really understand the change. But doing it as part of a huge redesign pollutes the data and makes it impossible to truly know why something is happening. You can make the metrics tell you whatever you want to see, there's too much noise and not enough clarity.