Hello Sir.  I am afraid.  Any grayscale support?

Quote from: graphmastur on July 24, 2010, 12:05:57 pm

I am already working on a client side program (Java, but obviously there are native files) which can access the usb.  It was designed to be used in conjunction with a battle network, or pokemon like trading system.  Using a real calc over a network is going to be slow as anything, there is no point to it.  If you want to program, do it on your own calc.  If you want to test compatibility, release it, and then get bug reports back.

Also, I don't know as you can access usb from python directly.  Anyway, I think it would be rather pointless as it would be horribly slow.  I personally don't like the idea of someone else messing with my calc, and I don't think others do as well.

Just my opinion though.  I can help you with usb a little if you need it.  Only the mac version is done so far. I am currently writing the java part around that.

Ahh, interesting. Your Java client is for the calc, right? And does it also have a corresponding calc program/app?

I personally don't like Java, since it's slow, and quite crazy to compile. Python is a lot easier.
PyUSB is a python library to access USB devices, and is cross platform. It uses libusb mostly I think.

But I won't know what to do anyway. My idea is for Quigbo to develop the calc side AND the PC side for the communication, then I'll port it back to Python and such. (Or he could just specify the way to contact the calc)

This idea hopefully isn't too bad - but it is kinda strange, so we'll see.
I think speed isn't too bad. The screenshots are streamed, and the screenshots are probably 100 KB max, maybe even less. It depends on server, PC, and calc speed.

And any help is appreciated!

I would be glad to help.  The front end is basically java, but I am using JNI as the back-end.  Eventually, I would like to write drivers and kexts to do away completely with the java program, and simply have the user have to plug the calc in and go.  All the abstraction would be done via the program on the calc.  The computer driver would just see that the calc was connected, and basically provide a pass-through to a server that the program specifies.  The java program is currently for testing the interface.

Are you aware of how usb devices work?  Please send me a PM, and I will help with whatever, since our ultimate goal is the same here.  I always had trouble with libusb.  Also, for a calc, all you really need is to deal with two pipes, so it's not all that hard.

Also, for what I do, Java is very easy, but I am also programming in C a little more.  As for Speed, I am thinking that the http is going to go faster than the calc.  Also, it would be impossible for grayscale.  Since my goals are for things like connecting a calc to a server anyway, I will help. (If you want me too.)

No problem.  This thread's other purpose, besides finding a new way to break RSA keys, is to explain how it works.  To explain modular arithmetic, it is basically this.

If you have 21 mod 5, that is the remainder of 21 divided by 5, or 1.  If you have 21 mod 4, that also equals 1 because the remainder of 21 divided by 4 equals 1.

Now 221 is congruent to 1 mod 55, because (221-1) mod 55=0.

Does that make sense?

Matthias, you are mistaken when you say it is one times mod.  Also, that's not an equal to symbol but a congruence.  (wikipedia)  basically, it means de-1 mod t=<equals>0, where t=<equals>(p-1)(q-1).  In other words, de-1 is evenly divisible by t.  Also look up modular multiplicative inverse on wikipedia.

Thank you so much. That is a lot easier to understand than wikipedia.  Also, I somewhat understand quadratic residues.

Oh yes.  That would be amazing.

Oh, I see.  What about the idea to use boinc to study the relationships between semiprimes and their factors?

Is it possible? (I mean I know that it is, but could WE do it?)

Maybe there is the software ability there, but not the actual hardware.  Eg, it could happen at some point but not at this time.