Hacker News Reader

Comment by eschutte2 | Hacker News Reader

I don't understand how that's possible. For example, in a recent project I needed to implement a path-finding algorithm to draw connections routed around arbitrarily-placed boxes. More recently, efficiently snap a line to the nearest edge from an in-memory set of (again, arbitrarily placed and changing) objects as the mouse moves. I'm having to use and think about some item from that list on a daily basis. Where's the disconnect here - are we talking about different kinds of front-end work?

View on HN

> Where's the disconnect here - are we talking about different kinds of front-end work?

Yeah I would assume so. Of course if know the job requires such knowledge, you should interview for it. But I'm thinking of more average front-end work, where you'd be building forms or dashboards or ordinary websites.

It sounds like you're building a complex application that just happens to be hosted in a browser. Although even then, I'd argue that while your knowledge of when to apply those algorithms is important, you still shouldn't be writing them yourself. You should be finding a well-vetted library and taking the function from there.

Who vets those libraries? And who writes them in the first place?

Not sure I understand the question... are you suggesting the concept of using vetted libraries is somehow faulty? Anyway, the answer is generally "a large community" or "some big company like google or facebook."

I think GP was referring to a situation where no such library exists or the needed modification would be prohibitive. Not that uncommon scenario once you are inventing new kinds of visualizations. Libs like d3js help, but they are pretty low level.

Those "large communities" and big companies are made up of people, who are front end developers are they not?

People knowing about algorithm+implementation details? Definitely not front-end developers. (Not trying to implicate that front-end devs can't do it... you get the point)

I don't really get the point. If you're writing an app to do X, and you require functionality Y for your interface, if a library doesn't exist, or is just some random persons GitHub dump minidress and no other committees, what do you do?

We may be on the same page here. I don't like the way dev interviews are run and I think the focus on textbook algorithms is misplaced. I would say if your interview process can be aced by somebody hitting the books for a month, you're not testing for hard enough things.

I was just a little surprised about those particular items because I feel like they're such an essential background for all software development. But I don't think they should be quizzed for in an interview.

So, so many people do things the hard and reliable way, or don't do them at all and sacrifice some desired feature, because they don't recognize a well understood computer science problem lurking underneath a business problem or a UI implementation.

I'm not saying that's what anyone in this discussion is doing, just that I see it everywhere (and have done it myself). 1000s of lines of incomprehensible code because no one realized that this drag and drop flow chart creation interface can be understood as a graph.

> I would say if your interview process can be aced by somebody hitting the books for a month, you're not testing for hard enough things.

I'd say if someone can independently learn the things you think are important enough to interview around in a month then they're the sort of person you should be hiring, unless you don't expect new problems to ever emerge you'll need people who can develop their knowledge anyway.

I think OP's list of essential algorithms is a bit too long. I agree that knowing what a binary search is, or graph traversal, is something that everyone writing code should know, and something that does come up reasonably often. (And recursion, heh.)

But binary trees? Last time I needed them was in a CS class. Linked lists? They only came up when I had to convince a coworker that we really don't need to implement one in JS, we only have 50 items here, and someone is going to forget to update a linked list if we use one.

This is why SV needs to stop abusing the term "engineer" - because this is indeed work that can be done on the front end, but under no circumstances should you expect front end developers to know how to do this. If you have a specific need, or you're building something very complex, then you probably want an engineer -- most companies (including any of the big ones) typically only need developers. It's good to have engineers tinkering with stuff though, that's how you wind up with things like React.

Right or you could have a "developer" tinker with stuff and that's how wind up with things like redux

Oh yeah absolutely - not trying to take anything away from developers particularly. Different skill sets, but practically the same outcome most of the time. That's a great example of why knowing how to create a BST from scratch isn't the only knowledge needed to build great software.

Most frontend jobs are gluing React, Ember, or Angular to the backend and sometimes making the CSS look right.

There ends up being a wide range of skills that fall into any classification and "front-end developer" is no different from what I see. For the majority of jobs, knowing html/css/js and modern js/css frameworks probably covers it. Your case sounds a little bit more specialized.

Then sometimes you see a "front-end developer" role and you find out it's actually a designer role.

I don't have a ton of actual front-end dev experience but when I get to do it in my consulting role I like it. So I'm trying to move into a role like that but the comments here make me think I don't have those skills. I don't have a strong CS background but was hoping I could still do front-end dev work. Then people are talking about algorithms and being a math minor and I'm not sure I could do that.

> I needed to implement a path-finding algorithm to draw connections routed around arbitrarily-placed boxes.

linked list, binary tree, dijkstra's algo - which one did you use? Why did you not settle with a O(n^3) or worse brute force?

Reply on news.ycombinator.com