« first day (2934 days earlier)      last day (1997 days later) » 
02:00 - 21:0021:00 - 00:00

9:00 PM
lots and lots
 
> They asked me if I had a degree in theoretical physics so I told them I had a theoretical degree in physics
or something like that
 
@danielunderwood Having stayed late in the office many times debugging something that (once) ended up being a strange edge case in the codebase where a string was being passed in where it should have been a list, I don't think Python is production grade. It is my preferred language for prototyping and can work for small projects and small teams, but anything beyond that I think you should choose a language where it itself enforces some sanity.
 
Yeah that's one of the things that's hopefully caught by automated testing, but python does leave it up to devs to not make mistakes in a lot of cases. There is some type assertion support to help that in newer versions, but I've rarely seen that used
 
so began the coding wars
how do you guys define the difference between scripting and coding?
Cus the guy also said Python was for scripting, it's not for coding. Scala is for coding.
Should we gatekeep coding outside of Python and other languages that dynamically allocate memory?
 
Do we agree that Bash is scripting?
 
9:14 PM
I dunno
 
I'd say scripting doesn't care about structure/performance/documentation/importable libs and is more for one-off tasks...though a lot of ops people work off a large collection of scripts that are organized and documented well. Then there's "game scripting" that's it's own thing
I'd certainly say that bash is scripting. I have seen some all-bash programs, but that's kind of out there
 
would you call bash coding?
I think the implication is that scripting and coding are exclusive
since the "not coding" modifier was using :P
 
I don't think I would call it coding
 
Would you call python coding?
or only scripting?
 
If you write on your CV that you know Bash or that you know Python (without elaborating more) vs you write that you know Java or C++, I do assume in the latter case that you know about structure, whereas the former case anything goes.
 
9:18 PM
I'd call 90% of my python coding, but I also do things that I'd call scripting such as a one-off script to process a file
 
I see
 
You can certainly write good structure with Python, but you don't need to think much at all to get something to work because it in no way forces you to.
 
@alarge I never thought of that, but that certainly is true
 
sounds fair
 
So if I write a PDE solver for my specific usecase where I know the boundary conditions etc etc, I'd write it in Python as a one off script. If I need it to work with a more general setup and thus do need to put more structure on it, I'd choose another language and it would need to be more rigorous.
 
9:20 PM
I don't think that's the case for me. If I needed to write something more general, I'd write it in python so I could make one-off scripts using it
 
sounds legit
I used to program in Fortran...I have no desire to go back to that one
C++ and Java are acceptable
 
And it is quite easy to make a dumb mistake in python, but in my case those have been acceptable things that hit an error reporting system and were fixed. If you were programming a pacemaker or rocket, it'd be a bit different. Though it's not necessarily any easier to make those mistakes than it is to get a bad memory access or memory leak in C++
 
the interpreter lets you get away with a lot more for sure
 
I guess Rust will in some years become the tool those people reach for where they used to go for C/C++
 
you won't run into any compile time errors...it's all run time errors...but there's a lot you can do to debug those run time errors
(apart from syntax errors)
 
9:28 PM
Yeah runtime errors are a bit annoying (you can even get a NameError during runtime...though that has some special reasons with how python handles variables). C++ doesn't check that you've allocated memory to that pointer you're trying to access either though. And C will happily let you access any bit of memory as whatever structure you're trying with
 
leading to segfault?
 
I think Rust could be ideal if we see its adoption. But most of the people coding at a level that needs rust have been using C/C++ for quite some time
The C++ would be a segfault. The C case could be a segfault or trash data
 
fun times
 
Or could be security holes.
 
I suppose some specialized systems may not even give you a segfault since I think that's up to the OS to say you can't access the memory
 
9:30 PM
extra fun times
I feel like that's all software engineering issues :P
I'm gonna just go ahead and deal with Data Science
if I work on a rocket or something, I'll make sure to let the software engineers know to be careful lol
 
NASA does have some special guidelines en.wikipedia.org/wiki/…
 
I guess I'll find some time at some point to study up more on C or C++
 
One of them is indeed heap allocation
 
I took a class in C++ in college that's about my exposure to that
looked over Java a bit
 
I tried to do some interview challenges in C++ and realized I was rustier than I thought
 
9:34 PM
Read a book by Meyers
 
Well not tried...I did them, but it was more of a chore than I thought
 
XD
 
Did you do interview challenges that were specifically C++ oriented and testing the knowledge of the language, or coded up general algorithm questions in C++?
 
Well it was a combination C++/Python thing that I could use either, but could choose what to use on each question. They didn't really test for language knowledge beyond knowing what was in the standard library
 
fun times
 
9:38 PM
It was more to determine that I could actually be given a problem and write out a solution rather than having in-depth knowledge of either language
 
I have in depth knowledge of nothing
not even my thesis topic
lawl
but I don't tell anyone else this
 
One interviewer asked me what I knew better than anyone else...I was just like uhhh
Or maybe that was a Dunning-Kruger test
On another note, do you use jupyter for your stuff at work?
 
nope
I code in sublime text
and run the code using anaconda prompt
technically I'm supposed to use visual studio for python
but.......for some reason it's super freaking slow for python...so I just use it to manage code sourcing through TFS
currently I'm the only person using TFS so it kind of defeats 99% of the purpose of TFS
lol
 
I use git even for personal projects that get a couple weeks of attention then ignored...not entirely useless I suppose
 
I mean, it's currently like a backup
 
9:48 PM
Do you commit on every model iteration or anything? Or even keep a record of old models that didn't work?
 
I commit when I feel like it
I'm a commitophobe
legit tho, I commit when I've written enough script to justify saving to an external source
like if I make some major-ish modification to the source code
as you can tell, I very much don't work like a legit software engineer lol
 
I think learning to commit at the right time is a bit of an art
 
it's an art that I don't anticipate having to learn too much...but maybe one day
my end goal is to get to a level where I don't actually have to code though lol
 
10:04 PM
What does that type of management role do? Would you still get to study models and just have other people implement them?
 
Well you can go do project management very quickly if that's what you want...
 
Sounds kind of boring not being hands on to me
 
not project management tho
like define the overall strategy
Director of A.I. type stuff
 
I think you should go do an MBA.
 
I think you should build AI that defines your strategy for you
Actually I think at a lower level, there are a couple companies that use AI for traditional project management. Don't really know much besides "someone is working on this" though
 
10:35 PM
dunno
I could do a MBA at some point I guess...
maybe a EMBA...XD
 
Do people do that? Have a PhD and go get an MBA?
 
MBA only 2 years, I don't see why it wouldn't be possible
but maybe a EMBA would be something I'd be more interested in doing...if I ever get that "E" XD
 
I've never even heard of that
Some secret club of the execs?
 
10:58 PM
Hmm
Where does the $2$ come from in the Stokes' stress constitutive equation $\tau = 2 \mu \epsilon$?
 
@danielunderwood an EMBA is for "execs" but it's mostly there because it's not a full time degree
it's more flexible so that you don't have to quit a job to pursue a degree...
 
11:11 PM
 
at some point I should clean up the codebase I wrote...
 
Don't a lot of master's programs let you go while working?
 
yeah...
I mean the EMBA is like a "lighter" degree
it's not as intensive I think as a full MBA
 
And I said that over a year ago about a project I wrote...it's still getting frequent updates with 10K users or so a day
 
maybe it focuses on some other stuff too
there's some arguments in functions that don't really need to be arguments and stuff like that...
 
11:20 PM
Read the negative comments on mba's
 
should be cleaned...but...I also don't want to break anything so I'm hesitant to clean it...
 
Yeah same situation here. The code is ugly, but there don't really seem to be any errors
 
yep...
 
@vzn what's the difference about getting excited about the Tenev stuff vs. the stuff in that article apart from the whole 'establishment' thing?
 
fight the man!
 
11:34 PM
Can't be part of the system!
 
rage against the dying of the light!
 
02:00 - 21:0021:00 - 00:00

« first day (2934 days earlier)      last day (1997 days later) »