Hacker News new | past | comments | ask | show | jobs | submit login
Codefights (codefights.com)
154 points by r_ms on Aug 21, 2014 | hide | past | web | favorite | 84 comments



I think you should show the comments first without a time-limit and then allow the user to play. For example:

        /*
         * @Input1: an integer
         * @Input2: an integer (@Input1 <= @Input2)
         * @Input3: an integer (@Input3 !== 0)
         * @Output: maximal integer from @Input1 to @Input2 inclusive
         * which is divisible by @Input3
         * or -1 if there in no such number
         */
<insert click to play button>

Right now it takes me 20-30 seconds just to read and understand the comments.


Why should reading the code be timed, but reading comments not timed?


Reading the comments is about understanding what the function is supposed to be doing. Reading the code is about actually finding the bug. I can see the argument for allowing one to understand the problem prior to hunting down the bug.


While I understand point people are making, I think understanding the problem is part of the debugging process, and should be a part of the puzzle and timer.

I remember entering programming competitions in high school and college. The first 5 minutes, when everyone is ripping open the envelope of their problem packet and trying to find the easiest problems was always very intense, and fun! Sometimes you would read through it too quickly, and miss a small detail that would render all the code you wrote completely useless. That's part of the challenge!


The difference in high school IMHO is that I give it a big chance that everybody was a native English speaker there.

I agree with you that understanding the (full) problem is part of the challenge in some cases, but since the headline here is 'How fast can you debug?' it has to test debugging skill.

I'm lucky to read and write English about every single day but have colleagues who are probably faster than me at debugging code that will take a lot longer because they'll first have to carefully read and translate the specification before they could continue.


I can definitely see both arguments. I think that they both have merits, and it's just a difference of opinion. :)


Because this is a test of ones ability to code; not the speed at which they can read English.

As I said earlier: https://news.ycombinator.com/item?id=8207116


One suggestion for the founders, don't start the timer until the code has loaded, and stop the timer once you submit. I lost 10-15 seconds to loading the code and then got game over because i submitted with 10 seconds left and it took longer than that to actually send.


Let's see how quickly they can fix that bug...


Yeah.. I got the task when the timer had 0:20 left, fixed it in 2 seconds, hit submit, it said "submitting..." until the timer ran down to 0:00 and then just "Game over". :-/


Yeap, great point. We currently sync the time with the server regularly to avoid cheating and given that the servers are under HN rush right now the delay is probably way more significant than it usually is. However, we are trying to figure out a better way to do this to make sure it doesn't start counting before the task loads.


Could you also start the timer after the player has acknowledged they've read the problem? At the moment you're unfairly disqualifying players who have reading difficulties (English isn't their first language, reading difficulties like dyslexia, or even those with partial eyesight so also read slower).

Timing the person's ability to read English doesn't prove their ability to write Javascript et al.


Won't people with dyslexia and poor eyesight have trouble reading the code as well?


Not with dyslexia, no. Because the problem lies with the syntax and semantics of the language and since programming languages are logic based rather than evolved from grunts with complex non-obvious rules, many people with learning difficulties (including myself) find it massively easier to parse code than English.

To use myself as an example, it took me around 30 seconds just to read and understand the first challenge, yet I was able to solve it in around 10 seconds after that despite it being a Javascript language (which is probably the weakest of the dozen or so languages I code in). If I was to go head to head with someone, that 30 second read time would be a hell of a handicap.

As for poorer eyesight, yes you'd have a point there, but at least the playing field would be levelled out a little more.


Interesting, I did not know that. I thought most of the problems with dyslexia came from actually being able to read series of letters and numbers - I've heard it described as letters coming together in a jumble, and not in actual order that they appear in a word


That can be an issue in more severe cases, but even in those instances most programming languages aren't too verbose in their syntax that you cannot understand the logic process. Plus indentation obviously helps a lot there too.


Same thing for me. Start the timer when the code is loaded and stop the timer when the user pressed Ctrl+Enter.

One other suggestion is to add Cmd+Enter for mac user.


Definitely an issue - I spent around 5 seconds between the code appearing and pressing the submit button, yet I ran out of time.


Hi guys, co-founder of codefights here, the site is still in beta and someone posting it to HN caught us totally off guard. Migrating the servers now to something that can handle this traffic. Appreciate the patience :)

EDIT: OK Servers migrated but it still can't come close to handling HN rush levels. While that's passing I can explain the main concept here.

A task on codefights is a function written in javascript that has 1 bug on 1 line of it and your goal is to find and fix it as soon as you can. There is a Solo mode and a VS Friend mode. In the Solo mode you start with easy tasks and need to find the bug in under a minute, if you do, you go to the next rounds where the tasks progressively get harder.

In the VS Friend mode, you codefight someone else where you guys have 5 rounds facing the same task in each round and each correct solution is 100 points, pass is 0 points and an incorrect answer is -25. At the end of 5 rounds whoever gets more points wins.

Please let me know if anyone has any questions or feedback. Hope you guys will get to try it for real soon :).


This looks really cool. Please consider adding a Github signup option.


Not needing signup would be even better :)


Thank Kenny, we are definitely considering adding a Github auth option.


Maybe I'm just showing my low IQ, but it takes me 20 or 30 seconds just to understand some of the problems. Maybe introduce difficulty levels? Currently, I can't ever solve a problem if it takes me more than 1 minute - I'd like to be able to get to an answer, even if it's then worth zero points (or even goes into negative points?).

I like the idea, but I can't play more than 2 or 3 rounds. Or maybe the game just isn't aimed for people like me...


FWIW, I have a high IQ and I didn't get very far either. Debugging code I've never seen before in less than 60 seconds isn't a skill I've needed to acquire.


I tried, I'm not a coder, submitted 2 of 3 solutions but just got "game over" back. Would be nice to see a solution, even better with explanation.

Definitely felt like I was being secretly profiled for a job. Also considered the possibility that if you're good they drop in some segments of code found automatically to have bugs and have you fix them for free ;0)


Dude, if you win it, the aliens come and take you away to hack in space. They've got some bugs that, if not solved, and quickly, could mean the end of galactic civilization.


That should be a good thing, understanding what you are about to fix should be most important. I don't think rushed fixes pan out too well in the real world.


I think the font size has something to do with it. Maybe the description should be in bigger font outside the code block?


I cannot connect with my Google account because I have an accent in my first name. And I cannot create an account with just my email (it sais "user not found")


I just gave up trying to connect it to Google or Twitter since those don't work. I ended up using the email signup option but neither signing up nor logging in happen over HTTPS.

That's why authentication through reliable big services like Google is a thing guys. At least they get it right.


And why no "Sign-in with Github" ?


How about just make it work?


Seriously. I just get a blank Google+ popup after authenticating. When I close the dialog it says it couldn't authenticate. Oh well. Looked like a cool idea.


same here.. I even tried using my valid email and all I got was "email is invalid"


It doesn't load at all for me... HN effect i guess.

But the idea looks like a very nice time waster or argument resolver.

I have a co-worker that always has the best theoretical cafe solution to every single bug ever found on any computer program, on this and parallel universes.

Would be nice to play this against him and maybe win one or 2.

Edit:

Looking at the bright side.

Most people that up voted the site liked the idea even without seeing the product working( I did).

That alone is a nice thing to have. So if you make it work you will get double XP!


Hehe, never thought of it as an argument resolver but that's a cool idea. One other idea that I have already tried quite successfully was to use this as a technical phone screen tool where you codefight the candidate or ask them to do a few solo modes to see how far they get :)


> One other idea that I have already tried quite successfully was to use this as a technical phone screen tool where you codefight the candidate or ask them to do a few solo modes to see how far they get

Sorry, but that's a terrible idea. As a bit of recreational fun, this game is fine for. But judging candidates on their ability to speed read would exclude a lot of very capable developers (as well as potentially land you into trouble under equal opportunity employment laws given the aforementioned bias this games introduces)


Interesting idea.

Several of the solutions I encountered are just... plain done wrong. They have a bug, sure, but the bug's actually there cause it was written in a really hacky way instead of the simplest way that would implement the function.

I feel like it took me longer to find the bug in the hacky implementation than it would have to rewrite it 'right'.

(For instance, the 'right' way to find "largest integer with x number of digits" is clearly and inarguably `Math.pow(10, x) - 1`. Not the hacky 4-line-with-for-loop thing they showed me with a bug in it. No?)

I wonder if it would be interesting to have something like this run your solution through a simple test suite to see if it works (like my university professors used to do on our homework), instead of accepting only one exact right answer.


This reminds me of more than one job interview I've had. One included a puzzler that was "fix this horribly convoluted javascript code by adding exactly four characters." I'd consider myself a pretty good bug finder-and-fixer. As in, I've read multiple books on debugging methods and even given some talks on effective debugging. Nevertheless, these kinds of stumpers make me freeze up.

I dunno exactly what puzzles like these demonstrate. That you spend a lot of time working in poorly written code?


I've actually seen this mentality on production code. "The smaller the diff the better. And The less we change the less likely we are to introduce new bugs!!1".


If you don't have tests, it's probably true.


The example on the homepage started a countdown from 1 minute, it was "Loading the task" until there were 7 seconds left, then by that time I'd lost interested and it just said "Game Over".

Game over indeed.


Hey Daniel, very sorry about that, HN rush hit our unprepared servers :) Just added several new servers so please try again.


Liking the idea, too bad it's simply an awful experience with all the HN traffic. I hit "Sign up with Facebook", then "Edit what CodeFights can see", and the counter had already begun. After authenticating I got an empty pop-up which I closed by myself, and I finished the challenge. I then got the error "Authentication error".

However, I like the idea and understand that this experience is not what you've made so I'll give it a chance and upvoted it.


thanks for understanding woutervdb and for the upvote, would be awesome if you can give it a try now (the HN traffic has calmed down :) ) and let us know what you think -- email info @ codefights.com


It should show you the correct answer if you get it wrong.


I agree. It's missing a learning experience here.


Absolutely agree. I find it a waste of time otherwise.


This is very nice. More engrossing than most games I play. Instead of starting over every time I lose, I would prefer if I was penalized and allowed to continue playing. For instance, if I lose round 5, then I will also lose all the points from round 4 and continue playing from round 4 instead. It's just something that I thought would make it more interesting and playable for me.


Seems really nice, but coul'd you add a solution if you fail in time? would be really awesome


From all the comments here I was expecting it wouldn't work at all without signing in somehow. Luckily that wasn't true and the site didn't seem slow at all so I got to actually try it.

I did like it but really wish there was an explanation at the end with the solution (or a solution). Since I don't know what is wrong I don't even know what to search for.

I am not sure if you have categories for the types of problems but dividing them into syntax/logic/easy/difficult/etc and showing your score on those or even compared to others (this might be what signing in does for you but I didn't do that).

I am sure I sound negative but I was impressed. Especially as it's still in progress.


Great idea but sadly I cant try it out :( Using Firefox on Android and the stock Browser I was finally able to login using g+ but I can't edit the code. I can add something but I can't delete anything that was already there. E.g. I can't change return a to return b. Using chrome I can't even login. Authenticate using g+ sometimes directly shows authorization failed and sometimes opens a new tab to codefights.com/_oauth/google?state=xyz&code=xyz that seems to do nothing. If I close it, I get authorization failed as before. Everything I currently do is done on Android 4.4.4 because I just have a smartphone (N5) available ATM.


I would love if HN had a "Save" button -- save an item into a personal list so I can, for instance, retry a link later after the HN effect has stopped...


Upvote the link, it will get saved under your profile "saved stories"


Yeah I know that but alas this isn't "Saved stories", this is barely "Random mess of stuff I've liked, stuff I wanted to keep an eye on and stuff I wanted to look at later"


Then write a browser extension to Tag these links locally.


Browser bookmarks?

I have a pile of "to do" "to read" "business idea" tags that I never get around to looking at ...


/ * @Input1: an array of strings * @Input2: a separator that will be used when concatinating * the strings from @Input1 * @Output: a concatinated string * @Example: myConcat(["Code","Fight","On","!"],"/")="Code/Fight/On/!/" */

Guess s/concatinated/concatenated/g wasn't the bug...


Is it anything like code challenges? I might add it to this list: http://codecondo.com/coding-challenges/


I am not at all sure about codeFights but tell you what.. the links on the page you mentioned are really good. I didn't know about a few of them and will definitely check them out. Thanks


After I fail a challenge, the screen disappears too quickly for me to continue looking at the challenge and figure out the correct solution. Is there a way to enable feedback or solutions?


there is actually a workaround right now (not a great one but still). If you go to Forum from the top nav bar then click tasks you can see all of the tasks that you ever got during your codefights.


I love the idea. It still needs more problems though. I kept hitting the regEx problem and failing because I've never done regEx in JavaScript. So adding more problems would be nice.


I think I'm fairly fast at debugging but these were way, way too hard with the strict time limit. It takes a lot of time to read the problems and the quality of the problems is poor.


Ghostery blocks something that seems to break the site altogether: http://i.imgur.com/hTkLTqd.png


I don't like how I can't change the number of lines of code. It'd be faster to completely rewrite some of the solutions!


Great to see CodeFights being launched! I enjoyed the graphics and dynamics of MathFights.com as well, well done, Tigran and the team.


Here's the problem: this is so awesome I forgot to come back and upvote until it was almost off the front page. THIS IS AWESOME!


the Oauth redirect receiver is failing for both Google and Twitter. (http://codefights.com/_oauth/twitter?close&state=*hashed_key...)


Twitter authorization doesn't work... unless that's the first round... did I fail?


1/2 the available time was taken waiting for "Loading the task..." !


sorry about that, way too many people on the site right now :(


This is a lot of fun - server is a bit laggy/buggy but unique idea, very cool!


Agreed -- it's unplayable as is, but it is an interesting concept.


This is a lot of fun.

Please say which language the problem is in.

Also, many times it seems to hang.


everything is in javascript right now


How do you verify answers? Say there is more than one correct answer to a solution: do you run the code in a sandbox (with memoization) to find out if it works properly? Or is there only supposed to be exactly one proper answer for each question?

EDIT You should also provide a "reset" button if I screw up the form.

EDIT The horizontal scrolling is messed up, Google Chrome 36.0.1985.143 (Official Build 287914) m on Windows. I cannot see the right two or so characters on a wide question.


This is like a hackathon project I had this year at HackRU.


I am still working on mine though, who made codefights? I would like to get in contact.


Sounds like Tiks is on the team that made it, commenting throughout.


I couldn't even try it. Waiting to check out tho


Hug of death?


is JS the only lang you support?


yes, at the moment at least




Registration is open for Startup School 2019. Classes start July 22nd.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: