What are you working on?

[Fourier]

[QUOTE=proboardslol;50131749]Won 3rd place at my community college stem fair for my modem project. Shout out to ma nigga fourier who helped me figure out how the fuck to measure frequency of a discrete time wave[/QUOTE]

gratz my fellow nigga! just keep it up!

[editline]14th April 2016[/editline]

[QUOTE=freakyy;50132222][B]K[/B]onsoleCommand?[/QUOTE]

He is Kroatian, mind that:)

[Map in a box]

[QUOTE=Fourier;50132269]He is Kroatian, mind that:)[/QUOTE]

I'm american, you don't see me spelling "color" like "colour" gosh

[Fourier]

[QUOTE=Janooba;50131205]I've never seen Part-Time programming work outside of contract work. (I haven't looked very hard though) I didn't think it existed.

Doesn't help that I'm in the least technical area of Canada as well...[/QUOTE]

I get part-time programming work requests from people all the fucking time. All sort of projects, web dev, virtual reality, electronics, unlock phone etc.

Maybe it's because I have built a reputation of godlike hacker - I did tons of projects in spare time, posted them on FB, blog etc, also every project people gave me ever I finished them, even if for free.

Maybe it's Also because I live in capital city of my country.

But start was tough, first project I was paid like 250€ per month, but it's true I did useless project (the one who hired me never profited off it).

[cartman300]

[QUOTE=freakyy;50132222][B]K[/B]onsoleCommand?[/QUOTE]

KDE Console is also called Konsole. It's called like that because there already is a default Console class.
I DO redirect the output of the console to my own thing tho, it's just you can't extend the default class with custom methods.

[editline]14th April 2016[/editline]

And having a Konsole class and ConsoleCommand and ConsoleVariable attributes is ugly and unintuitive.

[editline]14th April 2016[/editline]

[QUOTE=Fourier;50132269]
He is Kroatian, mind that:)[/QUOTE]

I do live in Croatia but i'm not Croatian.

[Tamschi]

[QUOTE=Protocol7;50131715]It's actually a pretty smart system if implemented well, it's the "implemented well" part that people tend to ignore.[/QUOTE]

There is one good part about it: You can easily GET and POST from JS running in a browser.

Otherwise you have:
- Massive overhead from HTTP (but that might be slightly less than with normal use if you have a specialised client, not a browser)
- Inefficient text encoding for everything
- [I]Multiple possible text encodings[/I], where it's difficult to support everything from scratch
- A delimited transfer format, meaning it's slow to parse or skip through
- Various undefined bits in the format, meaning it's possible to write two or more (defective) implementations that technically operate within the specification when talking to themselves but are otherwise incompatible
- Anything that's not GET or POST is inaccessible from a browser context (and most default HTTP clients) anyway, if I'm not mistaken.

It would make far more sense to write a decent generic binary protocol (that may negotiate some IDs based on strings as necessary), and to tunnel that through e.g. web sockets if there's really no option of opening a plain TCP socket.

[hypershadsy]

I've seen this a dozen times before on here, so I might as well ask:

I don't like using C# Console because it's limited in speed, color choices, etc. Does anyone have a library to simulate a console window in OpenGL or SFML or SDL or etc (like how Dwarf Fortress does it)? For bonus points, have it sit on SFML.Net.

[DFC]

I'm working on NPCs and pathfinding (writing my own so I can have NPCs walk around and pathfind with variable gravity, like this dude on his tiny red globe world). I think I've learned more about vectors in the past week than I have in all my years in school. I still need to figure out a good way of automatically placing the pathing nodes - as of now, the 'navmesh' is generated automatically but I have to place the nodes manually.

[thumb]http://cloud.screenpresso.com/1SQbc/globeworld_pathfinding.png[/thumb]

[cartman300]

[QUOTE=Hypershadsy;50133004]I've seen this a dozen times before on here, so I might as well ask:

I don't like using C# Console because it's limited in speed, color choices, etc. Does anyone have a library to simulate a console window in OpenGL or SFML or SDL or etc (like how Dwarf Fortress does it)? For bonus points, have it sit on SFML.Net.[/QUOTE]

[URL="https://github.com/cartman300/SFMLTextBuffer"]Text buffers[/URL], you have to manually manage scrolling tho, it's meant only as a drawing front end

[seefo]

[QUOTE=DFC;50133103]I'm working on NPCs and pathfinding (writing my own so I can have NPCs walk around and pathfind with variable gravity, like this dude on his tiny red globe world). I think I've learned more about vectors in the past week than I have in all my years in school. I still need to figure out a good way of automatically placing the pathing nodes - as of now, the 'navmesh' is generated automatically but I have to place the nodes manually.

[thumb]http://cloud.screenpresso.com/1SQbc/globeworld_pathfinding.png[/thumb][/QUOTE]

You can either place the nodes at every vertex of the mesh, or simplify the mesh (like for a sphere, where there are many vertices) similar to how LOD is handled and then place nodes at the vertices of the simplified mesh.

[Tamschi]

Codices and restoring instances work now:

// Console output

MemberFormat<Kodex_Test.Notable>(A, B, c, e.E, RestoreCount)
MemberNote<Kodex_Test.Notable>(A = ValueNode(1), B = ValueNode(2), c = ValueNode(c), e.E = ValueNode(5), RestoreCount = ValueNode(0))

Codex:
ValueNode(1)
ValueNode(2)
ValueNode(c)
ValueNode(5)
ValueNode(0)
MemberNote<Kodex_Test.Notable>(A = ValueNode(1), B = ValueNode(2), c = ValueNode(c), e.E = ValueNode(5), RestoreCount = ValueNode(0))
MemberNote<Kodex_Test.Notable>(A = ValueNode(1), B = ValueNode(2), c = ValueNode(c), e.E = ValueNode(5), RestoreCount = ValueNode(0))

First copy: MemberNote<Kodex_Test.Notable>(A = ValueNode(1), B = ValueNode(2), c = ValueNode(c), e.E = ValueNode(5), RestoreCount = ValueNode(1))
Second copy: MemberNote<Kodex_Test.Notable>(A = ValueNode(1), B = ValueNode(2), c = ValueNode(c), e.E = ValueNode(5), RestoreCount = ValueNode(2))

That mostly concludes the core library. There are still a few parts missing, but it works for most configurations now.

[awcmon]

like a year ago or some shit I think I was reading the CS:GO megathread on facepunch and somebody (im pretty sure it was BDA) wanted some sort of semi-auto script for tapshooting with rifles or some crap probably

with outside commitments and all, i haven't been programming in C++ for nearly half a year (java at school, c# at work). i recently returned to c++ when i made that one markov chain thing two or three pages back. i also returned to cs:go after like a year, and found out that i'm now terrible at the game

anyways i figured, i might as well write a program that does semi-automatic fire in CS:GO. it's a very small and simple program. here it is if anybody wants it. windows only.

#define WINVER 0x0500
#include <windows.h>
#include <iostream>

int main()
{
	INPUT ip;
	bool enabled = false;

	ip.type = INPUT_MOUSE;
	ip.mi.time = 0;
	ip.mi.dwExtraInfo = 0;
	ip.mi.dx = 0;
	ip.mi.dy = 0;
	while (1)
	{
		if (GetAsyncKeyState(VK_MBUTTON))
		{
			enabled = !enabled;

			ip.mi.dwFlags = MOUSEEVENTF_MIDDLEUP | MOUSEEVENTF_MOVE;
			SendInput(1, &ip, sizeof(INPUT));
		}

		if (enabled && GetAsyncKeyState(VK_LBUTTON))
		{
			Sleep(5);

			ip.mi.dwFlags = MOUSEEVENTF_LEFTUP | MOUSEEVENTF_MOVE;
			SendInput(1, &ip, sizeof(INPUT));
		}
	}

	return 0;
}

i'm pretty sure it won't get you VAC'd or anything. my autoclicker works using sendinput and i've used it for a long time and i never got banned, etc.

it isn't elegant or anything, but it gets the job done. middle mouse is to toggle between semi-auto and full-auto. i want the program to make a sound when i do so, but is there a simple way to do this short of using OpenAL or something? i don't care of the solution is windows only since my program is windows only already.

also, there is a weird bug where if you try to shoot pistols too quickly, sometimes it doesn't register. the Sleep(5); helps with it a bit, but might as well just disable the semi-auto script if you're using a pistol

chances are nobody here is actually going to use this, since most people here can probably just write their own. maybe BDA or some random dude 3 years from now might stumble upon it, idk.

[tisseman890]

I have managed to make a proof-of-concept native pagination for FP Droid

[vid]https://my.mixtape.moe/yhirkt.webm[/vid]

[Map in a box]

[QUOTE=Tamschi;50132992]There is one good part about it: You can easily GET and POST from JS running in a browser.

Otherwise you have:
- Massive overhead from HTTP (but that might be slightly less than with normal use if you have a specialised client, not a browser)
- Inefficient text encoding for everything
- [I]Multiple possible text encodings[/I], where it's difficult to support everything from scratch
- A delimited transfer format, meaning it's slow to parse or skip through
- Various undefined bits in the format, meaning it's possible to write two or more (defective) implementations that technically operate within the specification when talking to themselves but are otherwise incompatible
- Anything that's not GET or POST is inaccessible from a browser context (and most default HTTP clients) anyway, if I'm not mistaken.

It would make far more sense to write a decent generic binary protocol (that may negotiate some IDs based on strings as necessary), and to tunnel that through e.g. web sockets if there's really no option of opening a plain TCP socket.[/QUOTE]

!!actually!! since HTTP is so spread out, there are tons of load balancers everywhere that are smart and since they can understand the protocol, they can do a lot of predictive things that speed stuff up a bunch.
REST isn't just for browser usage, although for now(soon not to be iirc), JS can only get/post request so the method is simulated in a query parameter.
The spec is well defined enough that there aren't any confusions where there shouldn't be. Text encodings are very standardized and are easy to sort through.
HTTP has grown from what it was in the past.

[Tamschi]

[QUOTE=Map in a box;50135865]!!actually!! since HTTP is so spread out, there are tons of load balancers everywhere that are smart and since they can understand the protocol, they can do a lot of predictive things that speed stuff up a bunch.
REST isn't just for browser usage, although for now(soon not to be iirc), JS can only get/post request so the method is simulated in a query parameter.
The spec is well defined enough that there aren't any confusions where there shouldn't be. Text encodings are very standardized and are easy to sort through.
HTTP has grown from what it was in the past.[/QUOTE]

The load balancers embedded in the Internet infrastructure all work through DNS (multiple records) or on IP or TCP/UDP level, meaning TCP is equally affected.
For the endpoint load balancers, there's absolutely no reason why they wouldn't be able to use plain TCP (or web sockets), since that's fully under the control of the API provider and the TCP/binary load balancing would be (at least in theory) much more efficient due to vastly lower hardware performance requirements (since with HTTP you always have to scan through the whole message while with a good binary protocol you only need to examine very little of it and just forward the rest in larger chunks).

What [I]might[/I] make a difference is that you can in theory use existing cache providers to handle some of the GETs for you, but I'm pretty sure the otherwise inefficient protocol is quite the trade-off there.

[timvdlinde]

[QUOTE=tisseman890;50135833]I have managed to make a proof-of-concept native pagination for FP Droid

[vid]https://my.mixtape.moe/yhirkt.webm[/vid][/QUOTE]

Very Nice!!

[Funley]

I've been working on a mod manager for [URL="https://www.factorio.com/"]Factorio[/URL], that's more oriented towards making sure all your mods are fine and have no issues (for example, missing depended mods). It can sometimes even automatically fix said issues.

[t]http://i.imgur.com/zecqzIl.png[/t]

[t]http://i.imgur.com/sBvX6Cn.png[/t]

[t]http://i.imgur.com/ajJwWgV.png[/t]

The Factorio devs are adding their own native mod support inside the game, so this tool will be obsolete quite soon :( Still though, it's been useful for me so far and I've greatly enjoyed working on it. Even with the quirks of WPF.

[URL="https://github.com/Spanfile/Modtorio"]The repo in GitHub.[/URL]

[Number-41]

First holographic interferometry result!

[vid]https://dl.dropboxusercontent.com/u/17216535/ShareX/2016/04/2016-04-15_18-30-50.mp4[/vid]

In the left side you can sort of see a post-it note (rectangular structure). What I did was record an initial hologram (i.e. right before the brightness turns up) and add it to the real-time feed. this allows you to measure extremely fine phase differences. I didn't expect anything to move, but you can see the fringes creep up because the air around it is moving. The slight density variations might cause a phase shift. (I think, I actually have to ask my supervisor)

Next week I might post some standing waves in a circular membrane, they're pretty neat as well.

Also it's a 250 mW class 4 laser (a laser pointer is <1 mW) and it's creeping me out, it burns paper and even secondary reflections (i.e. looking at the dot on the wall) are dangerous :v: All the while I sit with my head at the same level as the beam so I cannot just casually turn around and look from my pc to my experiment...

[Fourier]

Hey man, this is some extremely interesting shit!

[editline]15th April 2016[/editline]

Why is this glowing square at center btw?

[Number-41]

Imagine the analogue holographic plate. Normally you would shine a laser through it and look from the other side. If you look at the other side, you would expect to see the image, but also a large part of the laser that just goes through it.

[img]http://i.imgur.com/U4qPCTH.png[/img]

In technical terms, it would be the zeroth order diffraction, i.e. the unaltered beam. Higher orders will make up the real and virtual image.

I highly suggest reading Gabor's (the inventor of holography) [URL="http://www.nobelprize.org/nobel_prizes/physics/laureates/1971/gabor-lecture.pdf"]Nobel prize lecture[/URL]. It is aimed at a wide public and it summarizes holography from its conception until 1970. The first two pages basically summarize the entire concept (it's fucking brilliant):

[QUOTE]After pondering this problem for a long time, a solution suddenly dawned on me, one fine day at Easter 1947, more or less as shown in Figure 2. Why not take a [I]bad[/I] electron picture, but one which contains the whole information, and correct it by optical means? It was clear to me for some time that this could be done, if at all, only with coherent electron beams, with electron waves which have a definite phase. But an ordinary photograph loses the phase completely, it records only the intensities. No wonder we lose the phase,if there is nothing to compare it with!

Let us see what happens if we add a standard to it, a “coherent background&#148;.My argument is illustrated in Figure 2, for the simple case when there is only one object point. The interference of the object wave and of the coherent background or “reference wave” will then produce interference fringes. There will be maxima wherever the phases of the two waves were identical.

Let us make a hard positive record, so that it transmits only at the maxima, and illuminate it with the reference source alone. Now the phases are of course right for the reference source A, but as at the slits the phases are identical, they must be right also for B; therefore the wave of B must also appear, reconstructed.[/QUOTE]

In my case, I digitally do the "shining of the laser through the holographic plate and looking from the other side". Usually one would crop out a quarter of the image.

[chimitos]

Laser goggles are cheap and very worth the investment. Blindspots are no fun. From experience.

Since you're probably using an IR diode to green laser, get a pair that catches both IR and green.

[vladka24]

Playing around with WPF trying out what I can do with it (pretty neat stuff):

[IMG]http://i.imgur.com/1gJ2FyN.png[/IMG]

[proboardslol]

I have to review a technical paper to save my CS grade. I was interested in Frequency Shift Keying, but every paper I find is more electrical engineering than it is CS. Does anyone have any suggestions fro interesting papers to read/review?

[larikang]

[QUOTE=proboardslol;50140034]I have to review a technical paper to save my CS grade. I was interested in Frequency Shift Keying, but every paper I find is more electrical engineering than it is CS. Does anyone have any suggestions fro interesting papers to read/review?[/QUOTE]
I just presented this paper for a seminar [url]http://www.arijuels.com/wp-content/uploads/2013/09/IEEE_SP2015_GenoGuard.pdf[/url]

Not very math-heavy, and pretty interesting if you can take the time to learn the genetic stuff.

[AntonioR]

Is there any good reason why would a game that comes in a simple rar archive save its config and save games in the AppData folder (and the Linux/Mac version of it), instead simply saving it where the exe is ?

[Geel]

[QUOTE=AntonioR;50140406]Is there any good reason why would a game that comes in a simple rar archive save its config and save games in the AppData folder (and the Linux/Mac version of it), instead simply saving it where the exe is ?[/QUOTE]

Consistency?

[Johnny on Flame]

[QUOTE=AntonioR;50140406]Is there any good reason why would a game that comes in a simple rar archive save its config and save games in the AppData folder (and the Linux/Mac version of it), instead simply saving it where the exe is ?[/QUOTE]

Security, you're not supposed to have write permissions to the applications installed unless you're the admin user, so you need to have a folder w/ rw access somewhere. appdata/home/whateverisinmac are used for consistency.

[roastchicken]

Except then you have games putting their data in Documents\My Games\, then another one in their own C:\ folder, then in Program Files...

Why couldn't Windows (and just OSes in general) be more specific as to save locations for program data.

[Johnny on Flame]

[QUOTE=roastchicken;50140562]Except then you have games putting their data in Documents\My Games\, then another one in their own C:\ folder, then in Program Files...

Why couldn't Windows (and just OSes in general) be more specific as to save locations for program data.[/QUOTE]

They are quite specific about their data, it's just that some developers aren't very bright.

[Map in a box]

My Games is the one windows wants IIRC. Of course, for Steam since they isolate all of their stuff, they have a cloud save folder in the steam dir for cloud saves.

[20k]

So I've been fiddling with my program which automatically works out interplanetary transfers, and I've updated it to be able to test an impulse at various velocities *and* angles, progressively refining the values and optimising for minimum distance between the probe and the destination planet

Turning your search space from n to n^2 is never a good plan, so I had to optimise the calculation a bit. Before, an n^2 gravity calculation (for 9 planets + sun + probe) was done every tick. Now, the course of the planets is precalculated once (rather than per tick), and the effect of the probe on the planets is ignored. This is not a massive speedup, but it turns the n^2 gravity check into a one time n^2 penalty, and n afterwards (ie like 5x faster overall, maybe more). Overall its about the same speed as it was before the introduction of the angle search

Its still not tremendously fast (1 minute for a particularly complex route eg gravity assists to neptune, 5 seconds for a simple one, I have to manually control the complexity, forcing the simulation to find more complex gravity assists by limiting the initial impulse of the probe), but the accuracy is quite improved. Some of the found courses involve eg multiple gravity assists (!) from jupiter to get to the destination. This is pretty cool as I'm giving it [i]no[/i] guidance whatsoever!

Here's a journey to neptune (ie really far away), with a gravity assist from jupiter. Its completely cool (at least for me, but I'm a big nerd). It takes something like 80 years in real time

[video=youtube;l2YEBb5kTyA]https://www.youtube.com/watch?v=l2YEBb5kTyA&feature=youtu.be[/video]

All the other planets are simulated, they just aren't displayed. The gravity assist from jupiter is about a minute in. The probes path after hitting neptune is probably inaccurate

The total search space for this is 400000 ticks * 10 angles per subdivision + 20 velocities per subdivision * 5 subdivisions. This makes it something like 400 million possibilities explored. There's probably a better optimisation technique here than brute force, but brute force is working for me at the moment