A more "real world" software development experience than what it might appear at first hand. Using a waterfall development process and referring to it as Agile. A team leader using their pet ideas and imposing their will on the other team members. Team leader ignoring input from other team members. Using a solution that doesn't fit the problem at hand guaranteeing a mediocre solution at best and at the same time greatly increasing the chances of a "train wreck" situation.

Accept this as a learning experience. When you graduate and are being interviewed, you will have more insight in determining the nature of the org that is interviewing you.

Saturday was the first time ever that Amazon sent me a wrong book. I got Linux Network Administrator's guide 3rd edition and ordered some book on mobile user interface design

Got the wrong guitar book once like two years ago

I disagree on needing real life domain experts ex post btw. We built software for a central bank and EADS (Airbus) with close to 0 real life expertise in either. First third of the projects is usually making sure you understand the processes 100%. Usually involves having a couple of people living with the customer and asking a lot of questions and learning the processes.
[we did study the domains before and built some generic processes that we thought would exist as quick prototypes for the contracting phase]

From my experience in ERP it's usually a question of "do we trust these people" above everything else. There's usually contracts in place that allow future customers to talk to existing reference customers to get a picture of how things were done, if there were issues etc. (once again 100% transparancy philosophy of my company, we didn't work for customers that didn't at least agree to talk to future customers on the phone for 30 minutes after we brokered the contact)

I think part of the problem is not really understanding what POS software does. In my mind though, if there's not some sort of "admin mode" that can change the menu and you hard code everything into it, what the **** is the point of the software, it basically becomes a thing with clicky buttons that can print out ****. oh wow, fun.

haha, yea, I had this thought. my real world job experience so far has me well prepared for this sort of thing and I remained professional with him, but I'm still really put off about how hostile he was towards ideas. Hard coding the buttons as sprites just sounds so AIDSy to me but oh well I'll program whatever he tells me to.

I am looking forward to getting more experience with C++ but it seems we're just using the totally wrong platform for this.

But anyway I could be way off but my idea was to have some sort of MenuItem class that kept a string as a name (which would be put on the button's label) and allow a user (admin) to change that at will, and in the final printOrder or sendOrder function or whatever we have, it would just take a vector of MenuItems as a parameter and print them accordingly. That's really not fleshed out at all but more along the lines of what I'm thinking. The button drawing and event stuff would be handled by a completely different part of the system. The bitch would be if you had say 14 menu items drawn on your menu screen and you wanted to change it to 12, it would have to resize and move some things around, which is what I thought he was worried about, so I told him we could just have a fixed number of items to save time/headaches.

But nope. sprites.

if in the middle of service the restaurant decides to change their tuna sandwiches to chicken sandwiches, I'm not sure how I'd deal with that, and I know I'm underestimating this, but it's really not a super complicated system at all compared to what people in this class have done in the past and is not something that should be causing panic attacks this early on about how to implement it ffs.

He asked me how I'd save a menu, I said I'd probably just save all the menu items in a .txt file and read that to populate a vector with menuItems or whatever the hell we decided on.

Obviously I have no idea what I'm doing but I'm here to learn so I'm irritated it gets shot down before it's even seriously talked about.

oh well, here we go

Hard-coded with sprites on a platform this dude made himself for video games definitely sounds like aids.

So, your proof of contradiction is to not hire domain experts but train them to be domain experts before letting them loose on the software phase?

I did suggest this possibility, but also said that it is much more expensive. Even if you don't have a genuine expert, it certainly helps to have a lead that can talk to customers on a level they both understand, being both personality and jargon.

I think that is one thing that makes me effective at my job. Most of my working career is manual labor, so I can talk to them at their level very easily, and I suppose I struggle with office-type people at times. Maybe this backfires as I'm looking for a new job.

Possibly related question: Companies working on driverless cars, do they just have software engineers who are also extremely good with physics and whatnot, or do they have good programmers talk to physicists to be able to figure out breaking force, safety stuff, etc?

Companies such as that just have very intelligent and fairly experienced people who are capable of coming up to speed really fast on whatever it takes.

It definitely takes a combination of skills, but it doesn't take people who already have the exact background required.

for example: http://www.wired.com/2012/01/facebook-server-lab/

A neighbor company building a recommendation algorithm hired a dude who had been working to try and find theoretical particles. The dude has since left and works at some online casino gaming company.

You just have to hire some of the brightest people you can find, and that takes an interesting challenge and really good pay.

There's enterprise software that requires domain expertise and there's enterprise software that doesn't require domain expertise - what's going on is that since your team/company/whatever lacks domain expertise, you only get to work on the latter.

The physics involved is not complex. It's not that difficult to build a physics simulation model because you just have to get the basics right and the rest can be learned from the actual world. It's much more about having a variety of sensors, a flexible model, crunching data and adjusting dynamically when expectations are proven wrong empirically than having a theoretically correct model. Physicists are generally only needed when you need novel physical theories or advanced topics that have to do with relativity or quantum effects - Newtonian physics is something that any layman smart enough to work on these projects can pick up quickly and something robotics people deal with routinely.

The experts they need are those on computer vision, machine learning (and not the kind of machine learning everyone else is talking about either) and various other AI and robotics topics. And you definitely don't need everyone to be an expert on everything either - I'm sure there's a lot of mundane work involved as well as some cutting edge research.

Edit: Btw, by not too difficult, I mean not too difficult from a physics perspective. I'm sure it's very challenging from a computational perspective especially when you have to deal with imperfect sensors, low-latency requirements and dynamic nature of the simulation - for example, tire pressures, car weight, road conditions, etc may all change even in the perfect circumstances and that's before you consider any malfunctioning. The challenges involved here have nothing to do with a high-level understanding of physics.

Someone on HN today claimed 95% of interview candidates could not write a function that returned the number of lowercase 'a' letters in a given string when given that question.. thoughts? By my estimate only 40% of people I've interviewed could pass fizzbuzz but the previous question seems more accessible/easy.

Unmeasurable and vague

Sounds like standard "I'm so smart and everyone else is an idiot" bullcrap complaining

I'd say that person totally blows at writing/posting job descriptions and/or screening resumes.

I'd say about half the people I interview can't do the programming questions we use in our phone screen well enough to move on. But probably only 15-20% totally can't do them at all.

I wonder how many javascript developers could do it without using a for loop/charAt. 5% sounds about right..

patio11 is sort of an HN celebrity. Not sure if he knows how to interview or not. I think the debate with tptacek boils down to the idea that no one knows how to interview.

Seems like most places in LA have this home-test / project style of pre-screening. I've managed to do fine on all of them, but I don't even think they are that effective. No one is going to spend more than a few hours to do this stuff, and the code isn't going to be stuff you'd insert into a valentines card to impress the hacker girl living down the street.

A wise man once told me that reg exps are the sign of the devil.

Uhhhhggh regex

In my advanced java class my proff ****ing love regex and it would frequently show up on exams, i hate it with a passion

Case in point, this is a thing:

^[+-]?(\d+\.?\d*|\.\d+)([eE][+-]?\d+)?$

If I don't use regex for about a month, I feel like I need to relearn it every time. Regex in exam feels pretty harsh, a better question imo would be ask you to match something in a string and give you a basic cheat sheet.

Regex should not be used outside of code golf imo.

Well, that's just not true. It's a tool like any other

Regex is great. Like any piece of "magic" though it gets overused in spots where people are using it to make ill advised shortcuts.

jmakin,

Why is that dude in charge anyway? Seems like he has no idea what he's doing.

Haven't even finished school and already getting job offers

From randoms on this forum

Wasn't thinking regex actually.