Programming

  On 12/5/2019 at 6:40 PM, Zeffolia said:

Definitely not true, there are many problems they can't solve well.  Drivers will never be automatically written using neural networks, nor cryptographic algorithms.  Non-NN AI though maybe.

True, I was exaggerating. But only a little. Most of the software that’s being written is fairly dumb business processing stuff. For that “AI” (with or without NN) will end up being good enough (and much cheaper than having code written by humans).

We’re fast moving in this direction. Case in point: Amazon CodeGuru.

https://aws.amazon.com/codeguru/

Do any of you guys have any recommendations for refactoring best practices?

I've never had to do it before, but I'm getting to a point with my scripts at work where it would be far better to have the experience.

https://refactoring.com
 

We teach this at my school. 
 

Alongside Robert Martin’s book “Clean Code”, in fact, which as it happens is not only a very good book but fun to read as well.

Thanks @rhmilo!

A book with "Clean Code" as its title... Please tell me it uses some form of C (++, *, regular) throughout the book. As there's some kind of poetic justice in using C to explain how to Code Clean.  

  On 12/10/2019 at 3:50 AM, goDel said:

A book with "Clean Code" as its title... Please tell me it uses some form of C (++, *, regular) throughout the book. As there's some kind of poetic justice in using C to explain how to Code Clean.  

Haha. No, it uses pseudo code that happens to be Java, except in one part where it actually is Java. Doesn’t matter, though, most principles hold true for all programming languages, even non-oo ones sometimes.

I usually code in emacs but have been really falling in love with spacemacs lately. Well, it runs on top of emacs, but anyway. This is some good shit. It also has a vim mode for you vim bastards out there.

Looks interesting. I may waste some time looking into this. Text editing in vim is far, far superior to anything else, but as an ide it’s quite lacking. That, and the idea of programming an editor with Lisp appeals to me immensely.

Update: @zkom : so thanks very much. Guess what I installed today ...

First impressions: vim emulation is really, really good. Much better than in IntelliJ, for example. It actually feels like vim itself. Weird: the terminal window can be edited using vim commands as well. Not sure what to think about that. What little IDE stuff I played around with is quite good. Probably not at the level of IntelliJ, but good enough for the coding I do these days, which is not a lot. 

The joke used to run that Emacs was a fine operating system that was only lacking a decent editor. From the looks of it, it's now got one.

Just came back to say: f. you @zkom for drawing my attention to this.

Spacemacs was the perfect gateway drug to get a taste of an editor with working vi keys that *can* do things like auto complete, spell check, auto fill without it feeling like it was hacked on with brittle tape adhesive. And org mode: oh my god that is fantastic.

It is, however, insanely slow to start up. Understanding everything it's doing is probably impossible. And for my minimalist-because-years-of-using-vim-makes-you-that-way tastes it attempts to do far, far too much. I really don't have any need for things like Helm and the gazillions of space bar based shortcuts.

So I have now spent at least three evenings (and who knows how much of my boss's time) creating my own Emacs config from scratch. Pretty fun little project and I now have an editor that does everything I need, can be easily extended if I decide I need more. Which I'm sure to do because, as it turns out, futzing with your editor config is really, really addictive.

Thank you for ruining my life

  On 1/17/2020 at 11:04 AM, rhmilo said:

Just came back to say: f. you @zkom for drawing my attention to this.

Spacemacs was the perfect gateway drug to get a taste of an editor with working vi keys that *can* do things like auto complete, spell check, auto fill without it feeling like it was hacked on with brittle tape adhesive. And org mode: oh my god that is fantastic.

It is, however, insanely slow to start up. Understanding everything it's doing is probably impossible. And for my minimalist-because-years-of-using-vim-makes-you-that-way tastes it attempts to do far, far too much. I really don't have any need for things like Helm and the gazillions of space bar based shortcuts.

So I have now spent at least three evenings (and who knows how much of my boss's time) creating my own Emacs config from scratch. Pretty fun little project and I now have an editor that does everything I need, can be easily extended if I decide I need more. Which I'm sure to do because, as it turns out, futzing with your editor config is really, really addictive.

Thank you for ruining my life

Expand  

You're welcome 

I just recommended spacemacs also to one of my colleagues at work who was using just vanilla emacs.

The software industry has an "interesting" feature that the skill levels of people working with software vary so wildly. And sometimes it seems like it doesn't even correlate that well on the years of experience. Sometimes the trainee working on his first job is much more skilled and productive than the "software architect" that has 20+ years of experience. Just blows my mind

  On 2/3/2020 at 11:33 AM, zkom said:

The software industry has an "interesting" feature that the skill levels of people working with software vary so wildly. And sometimes it seems like it doesn't even correlate that well on the years of experience. Sometimes the trainee working on his first job is much more skilled and productive than the "software architect" that has 20+ years of experience. Just blows my mind

Yeah. I see it in my students as well (I teach software development for a living). They all come in at roughly the same level and some of them just soar while others, soldierly and diligently, but nonetheless slowly, plod on.

Only thing these "Gods Walking Among Mere Mortals" (as their fellow students call them) are not very good at is planning their work, modelling their software beforehand and explaining their work afterwards, which unfortunately is an important skill to have when working in a corporate setting. They're also often, but not always, algorithmically weak.

  On 2/3/2020 at 2:10 PM, rhmilo said:

Only thing these "Gods Walking Among Mere Mortals" (as their fellow students call them) are not very good at is planning their work, modelling their software beforehand and explaining their work afterwards, which unfortunately is an important skill to have when working in a corporate setting. They're also often, but not always, algorithmically weak.

Having worked in some start-ups these kind of people sometimes gravitate to them, and they are really useful when coming up with the prototype or proof-of-concept but the productization then might cause problems when you need to actually document the whole thing, set up test automation and generally co-operate with others. F.e. there are some mathematically brilliant people that are doing software who have no idea how to work in a software project but are invaluable to the company.

Anyway, I also like it better to work in a startup than in a big corporation. Corporate jobs are cushy but can be so fucking tedious with all the bureaucracy and endless meetings. In my experience startup work is generally chaotic mess and more stressful but everything is very hands on and you don't need to ask much permissions to do things as long as it's taking the project forward.

  On 2/3/2020 at 3:19 PM, zkom said:

Having worked in some start-ups these kind of people sometimes gravitate to them, and they are really useful when coming up with the prototype or proof-of-concept but the productization then might cause problems when you need to actually document the whole thing, set up test automation and generally co-operate with others. F.e. there are some mathematically brilliant people that are doing software who have no idea how to work in a software project but are invaluable to the company.

Spot on.

  Quote

Anyway, I also like it better to work in a startup than in a big corporation. Corporate jobs are cushy but can be so fucking tedious with all the bureaucracy and endless meetings. In my experience startup work is generally chaotic mess and more stressful but everything is very hands on and you don't need to ask much permissions to do things as long as it's taking the project forward.

Never worked at a big corp, but at the startup type places I've worked after a while I got really, really tired of constantly having to fix the same crap that could've been prevented with just a little bit of the corporate thing.

And then when I started working with students, I quickly lost whatever disdain I had for corporate ways (quite a lot, at the time). When you're working with idiots, or rather, people who aren't very competent yet, all that up front design and documentation is god send. It may not reflect what's actually in the code, but at least it documents what it was originally intended to do.

So, yeah, testing is very useful. Separating out the design phase from the implementation phase is, too, because it makes your designs so much better (but it makes the coding itself boring). Only problem is it makes you easily replace, which is good for your boss but not quite so good for you. 

TBH, software engineering in itself isn't really my field but applied mathematics, which just usually ends up in writing a ton python code so it's basically a form programming. Because I don't normally write code directly for the product itself but for test and measurement automation and analysis instead the kind of strict quality control and following software development strategies isn't so important, just get the data out and analyzed in the best possible accuracy, reliability and efficiency. How the code looks like is secondary and sometimes anyway I'm the only developer and also the only user.

I did write embedded code for products in a huge-ass international corporation for four years and I can tell you never again. I also dipped my foot into the academic world and nope, not for me. The kind of niche I've found for myself is RF signal analysis and setting up test environments and automation for RF devices.

I  Golang

Paper on how they programmed the A.I. in F.E.A.R., considered to be one of the best A.I. in FPS gaming to date.

http://alumni.media.mit.edu/~jorkin/gdc2006_orkin_jeff_fear.pdf

^ Interesting and very accessible article