The article is somewhat incomplete as to how core memory works. Core memory was/is non volatile. Semiconductor memory hasn't been mainstream for that long in reality. I mean if we're talking 30-35 years as not being a long time. As the link I posted states, core dump is still a widely used term for example.

The pace of technological change is increasing in my view. The future will definitely be interesting.

Simplest way I can think of would be to create a socket between the node server and the c++ service that does all the heavy stuff.

If I were using mongodb in node, I probably would have the c++ service being able to connect as well and use the socket to pipe commands from node to c++ to then do whatever it is that needs done.

Socket would probably send commands by JSON with whatever else and c++ sends results back in JSON.

Another way off the top of my head would be to create a file using Node in some directory and the c++ service is scanning for when file is created and outputs the results to different directory, which node takes results.

I believe there is way to do it with streams in node with articles online.

Cool thing to note is that Socket.io 1.0 has binary streaming with sockets and they released a c++ example that I've used and its pretty cool.

Just woke up so post is mess lol.

Maybe I wasn't clear and put too much stuff in my posts. I'm specifically looking for stuff that isn't about bitcoin but more about blockchains in general (and high level). I think the abstract concept of having a "distributed ledger" can be used for other things than money.
Will read bitcoin related stuff as well though (Satoshi's paper is printed and ready to be read, he explains the chain pretty well).

My ramblings on JS weren't related to my thoughts on the blockchain but that library looks interesting regardless. I didn't consider calling stuff from node that sounds like a reasonable idea. My thought process was more along the lines of writing it in node then throwing it away if performance becomes an issue and rewriting in (pick your technology).

I haven't really followed what else the ledger can be used for besides property/money.

I've watched countless hour long talks on bitcoin(youtube).
The presenters did say it can bring other uses but i'm drawing a blank on what they are and i'm guessing nothing wild if i can't remember.

I think the stock market and trade/exchange could be done.
The ledger keeps track of trades and new companies are introduced as their own individual coin per se and still work on same ledger.
So you would have to rewrite the codebase but meh it would be cool as ****.

Thanks for the response. A socket interface would work. With sockets I would think of it as a client-server type pattern where the C-++ process would act in the server role. Have the C++ listen for requests to do the calculations and then serve them up. Nice. FWIW I would use blocking dockets but then I can think of reasons to use non blocking ones too. Cool idea. Hoping you'll post about your progress and what your findings are.

One prominent example (namecoin) is using that technology for a distributed dns of sorts. What I'm envisioning is some sort of ledger for supply chains (send x by y so we can produce z type of stuff) or transparent bookkeeping/accounting

Official document storage and signing is another use I've heard of.

Yeah it seems to get some attention from the legal folks. Also found a paper that discusses the applications for licence checking etc.

I pretty much want to get into a good brainstorming mode for interesting use cases. Between the blockchain, bittorrent synch like protocols and some stuff I've probably never heard about there has to be some exciting stuff one can create. I've pretty much put off thinking about any of this from a non consumer POV. Trying to fix that leak.

Just had a look at the cactus kev poker hand evaluator for the first time. How could one extend it to eval for straight draws? I've played a bit with low level stuff (leet hax0ring) but am pretty clueless especially when it comes to performance issues. Are precomputed hash tables basically always the best option? It's pretty obvious that if you or the rank bytes and have an oesd you'll have 4 consecutive 1's

xxxAKQJT 98765432
00001000 00000000
00000100 00000000
00000010 00000000
00000001 00000000
00000000 00000100

00001111 00000100

is there no quick way to just look for the presence of 1111 in the bytes? ie if I just do and's with
11110000 00000000
then
01111000 00000000
etc

edit: woops that wouldnt work... guess I'm asking if there is some fast way of doing this?


flush draw I think I figured out by anding and comparing with only 6 possible configs of two 1's and two 0's

CDHS
0100
0100
0010
0100
0100

0110

only care about 5 card hand evals for time being. sorry for sort of brain dumping

Pretty much what this thread is for, afaik

Studying Java this next year. About halfway thru the book we'll be using and I keep getting stuck on their style. I'm so used to putting an open bracket on the line beneath loops and whatnot so that all brackets line up.

Get really lost when it's at the end of the line. Pretty sure I spent more time trying to figure out scope for this sudoku solver they did than it took to understand what was going on. "This isn't going to accomplish anything because those variables go out of scope! ...oh wait, no, I see where those brackets are coming from now."

Not quite there, yet. Counter example:

CDHS
0100
0100
0010
0010
0100

0110

Same result, no flush draws.

Instinctively, I would work with sorting and then counting longest streak.

About the first part:
With the data representation you have chosen, you can skip 4 block comparisons if the last bit is zero. Or more generically, skip number of bits in comparison block minus the lowest order zero bit. The reason is simple, you need to start with at least one 1-bit to have a shot at finding an oesd.
That should cut down the amount of comparisons needed.

To give an example:

xxxAKQJT 98765432
00000000 00000010
00000000 00001000
00000010 00000000
00000000 00000001
00000000 00000100

00000010 00001111

Start at index 3 (0-2 aren't valid cards), compare 4 bits:
index 3, read 4: 0001: lowest order 0 bit position is 1 (counting from the rear, starting at 0); 4-1 = 3, skip 3 on index, new comparison index is 3+3=6. Got to index 6 and compare next 4 bits.
index 6, read 4: 1000: lowest order 0 bit is 0. 4-0 = 4, skip 4. New index is 6+4 = 10.
index 10, read 4: 0011: lowest order 0 bit is 2, 4-2 = 2, skip 2. New index is 10+2 = 12.
index 12, read 4: 1111. no 0 bit, found oesd.

Alternative exit. If at any time the starting index goes beyond total number of bits minus size of block (16-4 = 12), no oesd is available.

Again, that is just an option of how I would look at it to compare that against brute force comparisons. There is a solid chance that brute forcing is faster.

Just thought of another potential way. Just need to scan bits once.

Assuming you have those 16 bits in an array, here is some pseudo code.
At a maximum, we will loop through the bit array once. Early exits may shorten this.

ah... lol

I guess the method of assigning C,D,H,S prime numbers & using their product as a unique number should work the same as how cactus kev did with ranks. The lookup table would be pretty small.

Yea, I actually have a 7card nlhe range evaluator I wrote in ruby using all sorts of various sorting and bucketing techniques. It's fast enough to go through ranges pretty much instantaneously but I'm thinking of making one for PLO now and think my toy ruby program would take annoying long to go through a range. Seems like a good chance to play with some new things. Also why I'm looking for the fastest way to do things and keeping the sorting/looping/comparing to a minimum.

Heres how I did straights in ruby. It's the first thing I did in ruby so some things are pretty lol, didn't know snake case was ruby convention etc. Pretty sure it works though xD

http://pastebin.com/GaRTaVnT

Who's going to take one for the team?

There is no way to get rid of that icon I assume.

It is only a small icon I noticed a few hours ago. It may have been there for days for all I know. It isn't a splash-screen nag-box, so it doesn't bother me much.

"Mainstream" support for 7 ended back in January, but "extended" support ends in 2020, whatever that means.

http://windows.microsoft.com/en-us/windows/lifecycle

https://support.microsoft.com/en-us/gp/lifepolicy

I'll probably update to 10 if Parallels doesn't require a paid update to support it.
I use CLion on mac side instead of Visual Studio these days for C++ and VS was only reason I got Windows for my mac.

I'm really skeptical of Microsoft giving 10 free to everyone and I think they will start charging for security updates or something.
Yet OS X is now free and more people are discovering Linux distros. Maybe they are just going to make money from their office suite and be satisfied with that but only time will tell.

I clicked it, it's just to reserve copy of Win 10 on release which I did. My brother says the current Win 10 release very rusty, so a very ambitious launch date.

What about secure boot? In my view, FWIW, the only reason to have the option to disable secure boot is because Windows 7 isn't compatible with secure boot. I am sure that Microsoft would be more than happy to get rid of the option to disable secure boot. I realize several Linux distros have the appropriate keys that are compatible with secure boot but what about custom Linux builds that don't? I pretty much agree with comments Torvalds has made about secure boot, basically the cure is worse than the disease.

Create Your Own Keys for Booting Your Custom Linux Distros

This is a hassle. Pretty sure that not having the option to disable secure boot would complicate this already complicated process to greatly.

Secure boot is good and people that want it will always have it.
Not really sure why Microsoft would go out of their way to disable it.

What disease?

who uses wifi or sound these days? And launches programs by clicking on icons?

I have recently nuked my last windows VM because I read that our taxes can finally be paid via Linux (well some webportal of sorts not a native client). I have 0 reasons to need it outside of gaming which I don't do any more except for the occasional PS4 session.
Only reason I can ever see needing Windows again would be if I wanted to get back into poker. Since that was way too mentally taxing for me (I can't play casually, 100% or nothing...constantly thinking about hands etc.) I doubt that'll ever happen.

I'll stay with windows 7 until they pry it from my cold dead fingers

new laptop i just ordered has 8.1, not at all worried about moving up to 10, just so long as virtualBox works for it