Thanks, need the help to get more views.

There was the Interesting Stackoverflow links thread (which you actually started), but it sounds like you you were thinking of a thread to get more views to your question that needs answering.

I actually haven't asked that many questions. Usually my google-fu is good enough, but I can't seem to get an answer on this one.

Yea, I've seen some compare it to learning a new language and say it's not much of a problem to switch back and forth.

I'll give it a shot I think - the logic behind Dvorak makes a lot of sense to me. And if nothing else, it'll confuse and annoy people who try to use my computers which is always a bonus

Dvorak is so obviously superior to QWERTY, but remembering the pain of learning how to type on QWERTY stops me from even trying to learn Dvorak.

programmers.stackexchange has some really good discussions too.... usually more theory based.

Worked 3 days solid (12+ hour days) to write an online arcade.... all finished.... it works.... upload to live server and it's a great time (2am) to get an 'Unspecified Error' lol.

Any help?
http://stackoverflow.com/questions/8...ror-0xfffffffe

Just had another look and saw others feel the same way about the agents:

http://forums.aichallenge.org/viewtopic.php?f=21&t=1941

Juk

Hot damn finished!

Security concerns for people uploading HTML5 games to your web server was a real nightmare but I think we've solved it.

unhelpful when something the computer should be easily able to report returns "unknown error"!

Unknown errors/Unspecified errors are stupid

Here's another topic no one knows anything about:

I'm writing a relatively complex markov chaining program in PHP. The markov chain itself is not computationally expensive, but the outputs it gives must be then simulated at least 1 million times per event (and there's about 40 events per day). These events should be processed in batch and be ready a few hours after data is first sent to the program.

I've already written the program and the simulator works perfectly. However, as you may surmise, it runs pretty slowly. Basically there are finite state probabilities between 0 and 1 like so:

P(A): 0.000 - 0.110
P(B): 0.111 - 0.200
P(C): 0.201 - 0.300
...
...
P(Z): 0.998 - 1.000

(It's more than 4 significant digits in the floating point number but you get the idea)

The simulator calls mt_rand() to generate the probabilities, and for each of the 40 events per day, it takes about 50 of these probabilities. So 50 * 1,000,000 individual simulated states. Fortunately: They are not dependent on one another.

So, what seems like a good idea is writing it in parallel. The idea would be to give each child a set of rules (the P(...) rules) to process, and return the result sets for each probability, then sum them up.

Obviously this sounds like a perfect use of MapReduce, but I've only used it for MongoDB querying. I could also use pcntl_fork() in PHP but I've rarely worked with this.

Any suggestions on where to get started? All Googling has turned up are trivial examples that I've done but don't really help me understand how to do parallelized work with children returning a result set. A suggested PHP MapReduce library would be useful, or anything that uses pcntl_fork() - it can run on various Linux flavors.

haven't done PHP in years, so I can't help you with frameworks. But if the API you have doesn't seem to have a mechanism to return a value, there might be a way to initially pass in a reference to an array. Then when the job is done, you can access the results in the parent. If that doesn't work, you can always write to disk and read the results back in.

kyleb, why don't you pre-generate the random numbers? or extract the processing into a separate service? 1million requests is tough, but you could batch them into 100-1k, and have the service process them in parallel?

CTL+SHIFT?

What the ****? Windows is virtually unusable without that combo.

In all seriousness thought, that combo is an amazingly helpful hot-key. Taking that away would make me feel like someone tore my left hand off.

If you want to annoy people, just get one of these:

http://www.youtube.com/watch?v=vk6R3TzNzM4

All I know is that I really want one.

eh, it might be CTRL + ALT or something? it's a chord i hit every once in a while on accident whatever it is, though tbf i'm twitchy on the keyboard and/or mouse when i'm not actually typing.

I just finished problem 315 on Project Euler and I am pretty happy with myself.

Finding ' int __builtin_popcount (unsigned int x); ' made it much easier.

The function takes an integer, looks at it in binary representation and returns the amount of one's that are on. The best part is it does it in one machine operation[1].

[1]According to http://gurmeet.net/puzzles/fast-bit-counting-routines/

I just did my first 2 project Eulers:

Apparently sum n = n(n+1)/2 is "widely known."

I had no idea....

Re: this

Is it possible with the program to "click" on links in the URL and copy that as well?
This project is driving me nuts, I need to learn a real programming language.

Just imagine the combination of that and Dvorak. (making that vid 7 minutes long was a bit overkill though)


It's ALT+SHIFT in windows.

Even if you didn't know it then there is an easy way to see how the formula is derived:

If you visualize a grid of of size n x n and then cut it in half diagonally, but keeping the diagonal, like so:

XOO
XXO
XXX

or

XOOO
XXOO
XXXO
XXXX

or

XOOOO
XXOOO
XXXOO
XXXXO
XXXXX

You can see the formula for the X's is: (n^2 + n) / 2

Where, the n^2 part generates the whole grid:

XXXX
XXXX
XXXX
XXXX

the + n part duplicates the diagonal:

XXXXX
XXXXX
XXXXX
XXXXX

and the / 2 part gives you what you want:

XXXXX
XXXXX
XXXXX
XXXXX

The (n^2 + n) expression is the same as n(n+1) in your formula.

You can also see from the pictures why these are called "triangular numbers".

Juk

I don't think so - I tried using it to make a mirror of the overcards.com site and wanted it to do the same (ie: click stuff to get the site to execute it's php), but it didn't do as hoped/wanted.

Juk

There has to be a program that does it. I am not going to click on 2000 links and copy paste it. FML!

Who do I have to call or sell my soul to to get the program I need?

I need a software or code that lets me open 200 websites. I got a javascript one from another site, but since I am illiterate in JS, I cant modify it.

I want the script to open the websites in a new tab instead of a small tiny unusable window. And I also dont want the site refreshed or anything, I need to click on it afterwards.

Ignore the first website and use a different one.

Aren't there progs that download web pages for off line reading that will follow links?

are you trying to solve this programmatically or do you just want the website backed up locally? if the latter, and you dont want to use the program listed, just download/use wget

HTTrack will follow links deep in to the site. I scraped a website to 3 levels of navigation (all it had) with it. Just dig through the settings, it's all in there.