@Quincunx Looks like a boot on a rope

hmmm

I see Gandalf wearing an apron.

@Quincunx I thought it was due to the early history of USPS. After the Pony Express, they started using religious organizations since there were churches in most towns. Hence, the nun.

A boot launched from a giant slingshot*

@Geobits That reminds me of something my Grandad says about WWII.

Neither he nor the other men in his battalion spoke French, but he'd been to a good school and spoke Latin. So whenever they arrived at a town, they would find the local priest and speak to him in Latin.

@Geobits Apparently, there is this, but it doesn't seem to be in the standard library for some reason. Gotta find out how portable that is.

Good thinking. Any time I needed something in Japan that couldn't be explained with hand motions and my limited Japanese, a local Starbucks was perfect. They tend to hire mostly bilingual there.

@Rusher Spawning a thread, are you nuts? Yeah no, it's slightly above that curve for now. I'd much prefer simply non-blocking IO.

@m.buettner Yeah, looks like just the thing (if it's portable).

I prefer wibbley-wobbley threadey-mutexey stuff over nonblocking IO

@Geobits Ah I might just roll my own (read that as "steal a snippet from StackOverflow")

@mniip My multi-threaded programming experience in any language is rather limited (like I know the concepts and some theory, but I've never actually used it, apart from annotating a loop in C with #pragma omp parallel for), so I don't really want to look into it for Ruby right now (which I'm not that familiar with anyway)

Are you implementing this in Ruby just for the challenge? It seems like it'd be easier with a language you're familiar with ;)

I've threaded in C, C++, python, and Lua

@Geobits The languages I'm really comfortable with seem rather unsuitable for a code-challenge controller. Ruby seems to be very convenient for the large part of the code, and I'd really like to learn more Ruby anyway.

or I'll just wrap the entire game loop in this: ruby-doc.org/stdlib-2.1.1/libdoc/timeout/rdoc/Timeout.html

So I need an opinion here (ideally from KotH hosters and participants, both). I need some form of time out, to make sure that solutions scale linearly (well and to ensure that I ever get a result when benchmarking solutions). Since this is a game with multiple moves, where the communication goes back and forth between controller and bot, would you prefer:

To be told the time limit at the beginning of the game and keep track yourself.

To be told the remaining time at each single turn.

Not care because you wouldn't use it in your implementation anyway. (and instead just make sure that your algorithm runs in reasonable time for the given benchmark tests)

So you're using a game-timeout instead of a turn-timeout?

Yeah, that's what I was thinking, so you can take a minute or so to plan your race, but not have that minute for every single turn.

I'm open to all sorts of suggestions here, really.

I was even thinking about using a mix. You have a game time-out, but time from that is only deducted if your turn exceeds a certain turn time-out.

Hmm. If you're doing it that way, you probably need to give remaining time each turn. There could be a bit of a difference each turn between what I clock and what you clock, and those would add up each turn.

The advantage would be that you wouldn't have to worry about IO or stuff slowing down the playout and you not being able to finish the game after planning it all out nicely.

I like the idea of a game clock, though. It would make it beneficial to estimate how many turns you'll need early on, so you can cut yourself off if you're exceeding the average.

@Geobits still, if there's only a game clock, you might think you can still finish the game by making quick or precomputed moves, but end up timing out because suddely IO hangs for a bit, or you just misjudged the average move time you can do

If IO hangs, all bets are off no matter what clock you use, no?

Misjudging is your own fault, though.

@Geobits if it hangs badly, yes

What do you think about the hybrid case? Say the maximum number of turns is 60. You get a budget of 60 seconds, but time will only be taken from that budget, if any single turn takes longer than one second.

Quickly gotta head out for 20-30 minutes or so.

Discuss! ;)

That sounds reasonable.

Is it going overboard to make something like this for people who want to participate in my challenge but don't know exactly what does to stdout? docs.google.com/spreadsheets/d/…

That document is open edit btw, so add to your heart's content

And now everyone knows my name :)

An anonymous duck has joined.

It does feel a bit overboard. Just sayin.

Considering probably 50% of the questions on the site mention stdin/out, without the need to break it down 101-style.

I'm just thinking that 90 people starred the wolf question and I got 50 answers. What happened to the other 40? I am lowering the learning curve for them.

I'll just trash it. They can ask if they don't know.

Please?

Please what?

Please, sir?

yes ?

Please the sir

@mniip you really do prefer threading and stuff to async I/O ? :)

Yes

I had a nightmare once,

where I was running away from select()

well select does suck, that's a given

still, yours is an interesting opinion, even though I disagree, especially after having started programming in Go, which hides async I/O behind goroutines

Wait

are we talking async IO or non-blocking IO

I see the former as the usual implementation for the later

uh?

for me it's 2 not really related conceptrs

@mniip I beg to differ, but please enlight me with your definition for each concept

Async IO - registering callbacks that would be called when an IO operation finishes

Non-blocking IO - when an IO function returns its unability to do something, instead of blocking until it can.

thanks, I see what you mean, now

@Rusher to me, the more important gotcha is that you've got to remember to flush your out-buffer in many languages.

Also, I tend to star challenges even if I never had any intention to participate (not because of difficulty or anything, just in general), but where I'm interested in seeing what people come up with

Likewise, I rarely star questions where I'm writing an answer, because I know I can find it in my answered questions

@m.buettner You obviously haven't answered enough questions ;)

@PeterTaylor I don't usually look through my favourited questions unless I am specifically looking for something. If I am specifically looking for something, I usually remember if I answered that question or not. If I answered it, I'm quicker searching for it than browsing through my favourites anyway.

It sometimes takes me a few minutes to find an answer which I know I wrote once

Anyway I'm not searching for old answers of mine often anyway (especially on code golf).

I typically use favourites for questions which I intend to answer in the future.

But I do sometimes want to find questions which I've answered in the past because I'm looking for a dupe and not finding it.

Fair enough.

@Rusher Have you done the maths for your KotH? If you get 30 submissions, you'll be running the bots for up to a full day, if I'm not mistaken.

@Geobits @PeterTaylor I added some changes regarding the time limit: meta.codegolf.stackexchange.com/a/1695/8478 Is that better now? I'm still slightly worried what to do if anyone manages to find an optimal algorithm that runs in O(n²).

@Rusher Please destroy all teh sandboxes! ;P

@hichris123 In favour of...?

Using the regular SE system of closing, editing, and commenting?

Hm, fair enough. I think it leads to more frustration though and it doesn't seem viable if PPCG ever becomes active enough such that sorting by active questions gets unbearable (if you're sorting by new questions, you won't notice any more that any question made it past the sandboxing stage).