@whatsisname so it's a boolean flip rather than a gradual oscillation?

wired.com/2015/11/null

@durron597 decoration was all I could come up with for the ones I've seen... they're always super stiff to hold that perfect-round shape which makes them no good for anything but tossing on the floor that I figure. Though now that I think about it, if it's extra long I can imagine one plausible intent: Bed divider for people uncomfortable sharing a bed? Maybe? No, it's just a useless decoration; you might as well be asking how to make tea with potpourri.

@LightnessRacesinOrbit I use it on all my computers. On my work computer I run it during the day to avoid eye strain

white backgrounds (especially for text editors) are really intensely hard on eyes

when you work in them 8+ hours aday

@Snowman your problem is not enough scotch. More scotch + leaking gas appliance = no more leaking gas appliance. Scotch solves all problems, sometimes you just aren't using enough of it is all.

would it be correct to say that "failure atomicity" is the Java term for what C++ coders call the "strong exception guarantee"?

> Generally speaking, a failed method invocation should leave the object in the state that it was in prior to the invocation. A method with this property is said to be failure atomic.

Ugh, this coding style is a crime against humanity.

probably part of the reason he can't see the bug

There's something about this generated avatar that I find offensive :|

looks fine to me

think of the children.

I assume he was going a non-eye-melting bracing style that still had one or two elements of the "style" OP originally used

Hey, guys I'm looking for a free software project to get involved in. I'm officially a math/physics major but I know some CS stuff as well. I'm looking for something that has both academic and industrial merit. An example, I guess would be something like OpenCV. Can you guys suggest anything?

OpenCV?

Here's one approach - take something that you're interested in and use, think of how it could be better, and submit a patch.

Another approach: read an interesting lib's code base from one end to the other, and try to follow from entry to exit, while improving documentaton. Submit a patch.

@picaposo ^^^^

@MetaFight Bionic Commando was freaking awesome! (did we already have this conversation?)

@AaronHall documentation patches are So. Incredibly. Valuable. Most devs simply forget thorough docs because they already know the whole system, and have no idea what would be useful for a newcomer.

I've never seen docs that couldn't be improved in one way or another – though many of the hipper language's ecosystems manage to hit new lows … as in, no docs at all :(

YAGNI I think

I guess they both work

I rarely write documentation but that's mostly because I rarely write code that's intended to be used by other teams

@amon What would be useful to a newcomer: For each (library function) { what it does; short code example; }

YAGUI - new one to me

especially the code example, just yesterday there was a method I needed to use and the prose documentation made sense, but I had no idea what shape the argument was supposed to be or where I was meant to get it from

@RobertHarvey From excellent reference docs, I expect a precise type signature (esp. in dynamic languages!); a short description; description, types, and default values of parameters; behaviour on errors; discussion of special cases if any; simple example usage. An {example, type sig, description} alone is insufficient. Sadly, many libs only have an insufficient tutoral full of unusable examples.

Good documentation is time consuming, and hard if you don't have the skills.

I can understand missing some of that stuff, but anytime you create a public-facing method you must have put a bit of thought into its name, parameters, intended use case and all that; not taking the ten seconds to write down what you already know is just silly

The best documentation I ever wrote: a series of PowerPoints. Slide after slide of screenshots: do this, click that, move over here. I say the best because there was evidence that it actually got used.

I should really go write proper README.md files for the reusable modules I've put out there

I guess I already did that for the handful that I originally wrote with reusability in mind

@Ixrec in my experience, implementation, documentation, and unit tests take about the same amount of time. Because they contain the same info in different representations. But in fact, docs and tests are much more involved because they force you to think everything through carefully. Cowboy coding is much easier.

but I know there's some stuff in that repo which could use better docs

@amon I'm thinking of the straightforward kind of per-field/per-method documentation

and tests should be a net time saver if done properly because they catch bugs much sooner and make it far easier to do significant refactors without wasting tons of time testing manually

at least the ones I've gone to the trouble of writing so far have saved me tremendous amounts of time later on

I recently lost an hour debugging a failing test, only to realize that the test was wrong :/ But I can't imagine being able to create correct code without tests. Nontrivial programming is just too complex for that.

if the code's sufficiently modular it can work without tests, but everything is just better when test are there (speaking from experience, I put a lot of effort into working out how to test our frontend-ish code properly and am so glad I did)

@amon if the test turns out to be wrong, normally that should either be immediately obvious, or it indicates the test was not as readable as it should be, right?

@Ixrec it was a copy&paste error … and after seeing it, it was immediately obvious. → never copy&paste when programming, always re-type.

I copy and paste all the time, though usually because I want to duplicate the shape of something and only change names or add a few lines inside, without retyping all the keywords and indentation and stuff

though I hear most programmers rely on IDEs with magical autocompletion, unlike me