next »

[qMopey]

Yep STL is garbage if you like using debug mode.

[Ordnas]

Maybe it is intended by Microsoft, so you don't forget to profile in Release 

[Guntha]

That becomes a problem when you create tools intended for debug use only... I had created one such tool, that relied heavily on std::vector and made the program running too slow in a lot of cases, and just getting rid of most operator[] uses turned it almost in release-mode like performance :p

[FrankForce]

I am happy today because I made a flappy bird game on dwitter (140 characters of javascript)!

https://www.dwitter.net/d/13750

[FrankForce]

A "quine" is a program that print's out it's own code. After programming for 20 years, I finally got around to making one, and it fits in a tweet!

https://www.dwitter.net/d/13794

Code:

Z=2e3;for(i=0;i<Z;x.fillRect((S(i)*1e9)%Z,i,j=i++?i%7:Z,j))x.fillStyle=R(G=i*255,G,G);x.font=4+"em'";x.textAlign='center';code.split(Q='\x3B').forEach((s,i)=>x.fillText(s+Q,960,1140+89*(i-t)))

[Schrompf]

Hehe, nice!

[oahda]

Cool!

[Eggy]

Ok so as a non-programmer who fiddled with programming in the past (I am more of an artist), I've decided to take upon some programming again after being inspired by another person's work. Basically I'd like to give Monogame a try, so I am practicing some basics in C# via tutorials. So far it's been going pretty well and my first achievement is this:

And there's a lot more tutorials waiting for me to go through plus the potential to experiment on my own once I've mastered the basics, and the fact I'm learning coding for real feels amazing. I really have no clue why I haven't truly tackled something like this sooner.

[oahda]

Nice to hear that you're enjoying it! Good luck in making whatever it is you've set out to make!

[Ordnas]

That tutorial reminds me of a code test I made at the University (write on console * in a pyramid way), the difference was that the language was Java 
Good luck with you progress! 

[Eggy]

Thanks! I constantly imagine myself creating like an RPG or something but at this time that one's nothing more than a pipe dream lmao. So right now I'm starting small and trying not to rush into amazingly complex programming right away without first taking everything in and actually understanding the code I want/need to write
I think I'm advancing pretty well with C#. Currently on a tutorial that teaches me how to create functions/methods with classes waiting right after. Can't wait to get to the part where I'll be able to create an actual game.

[Ordnas]

Actually I also imagine myself gaining exp level like in a RPG when I started programming 
Good that you start slowly, it is like you can't start a marathon if you didn't learn to walk.
I think that you can also start with Unity instead of waiting to create a game one day. For example you can divide your day time like 30min with C# coding and other 30min make a game in Unity and C#.

[ThemsAllTook]

This is one of those posts where I was super grumpy for a while, but finally got to flip around to being happy when everything starts working.

For a while, I haven't been quite satisfied with my choice of audio APIs. Years ago I used OpenAL, but it was way too high-level and didn't give me the control I needed for a specific effect. I switched to portaudio, which was pretty great for a while - instead of using the API to manage audio channels and sources and stuff like that, it just gives me a callback function that I implement to fill a buffer with arbitrary sample data to be sent to the speakers. Huge improvement.

Lately I've been having some trouble getting portaudio to compile and behave correctly on all the platforms I want to support, and I'm always looking for ways to reduce third-party dependencies. I decided to implement my own bare bones audio output API which would work with the same sort of output callback system, but on the inside would directly use the native audio APIs of each platform I'm supporting without an extra layer in between.

I started with Core Audio on OS X, and man oh man, it's a complete disaster to get started with it. Documentation basically doesn't exist. I couldn't find any official sample code, and a cursory Google search didn't turn up any high-quality unofficial examples either. Looking at the source for portaudio and other open source projects that use Core Audio was moderately informative, but the APIs they were using are both enormously complex and deprecated. Apple's comments about deprecated functions only gave the vaguest hints about what to use as replacements, and without actual documentation or examples, I was left guessing at which parts were even necessary for what I wanted to do.

However, after spending enough time reading Core Audio headers and making educated guesses, I managed to get it to work with this tiny implementation:

Code:

#include <CoreAudio/AudioHardware.h>

static AudioOutCallback outputCallback;
static AudioObjectID device;
static AudioDeviceIOProcID ioProcID;
static AudioStreamBasicDescription streamDescription;

static OSStatus ioProc(AudioObjectID inDevice, const AudioTimeStamp * inNow, const AudioBufferList * inInputData, const AudioTimeStamp * inInputTime, AudioBufferList * outOutputData, const AudioTimeStamp * inOutputTime, void * __nullable inClientData) {
outputCallback(outOutputData->mBuffers[0].mData,
               outOutputData->mBuffers[0].mDataByteSize / streamDescription.mBytesPerFrame,
               inClientData);
}

void AudioOut_init() {
AudioObjectPropertyAddress objectAddress = {kAudioHardwarePropertyDefaultOutputDevice, kAudioObjectPropertyScopeGlobal, kAudioObjectPropertyElementMaster};
AudioObjectPropertyAddress descriptionAddress = {kAudioStreamPropertyVirtualFormat, kAudioObjectPropertyScopeOutput, kAudioObjectPropertyElementMaster};
OSStatus status;
UInt32 size;

size = sizeof(device);
AudioObjectGetPropertyData(kAudioObjectSystemObject, &objectAddress, 0, NULL, &size, &device);
size = sizeof(streamDescription);
AudioObjectGetPropertyData(device, &descriptionAddress, 0, NULL, &size, &streamDescription);
}

void AudioOut_startOutput(AudioOutCallback callback, void * context) {
outputCallback = callback;
AudioDeviceCreateIOProcID(device, ioProc, context, &ioProcID);
AudioDeviceStart(device, ioProcID);
}

That's it. That's the whole thing, minus a small amount of error handling. Based on all of the other Core Audio code I'd seen, I was worried this would be pages and pages of audio units, device management, and other nonsense. It's nice when an API gets this much better. If they would just create some actual documentation and sample code, everything would be perfect.

[qMopey]

I had a similar problem with Core Audio. I had to read docs from an old book on Core Audio I found on Amazon.com. Why not give cute sound a try? It does exactly what you’ve just stated you’re trying to do. https://github.com/RandyGaul/cute_headers

[ThemsAllTook]

Oh nice! That'll be a useful example when I go to implement the glue for other platform APIs. Thanks for the link.

[qMopey]

No problem! It’s a great reference for OS level audio APIs (except Linux where I just used SDL).

[oahda]

Got AngelScript up and running in my engine and it’s really nice to be able to have a virtually identical API to access in both the parts of the game written in C++ and those written in AS—don’t have to learn a secondary system for the scripts and jump mentally back and forth between the two depending on what I’m working on. 

[Schrompf]

Have been using AngelScript for a literal decade now. As long as you're the only dev, it's great. It has its downsides, like the convoluted creation/copying flags you have to guess for each new class, but it got better over the years - now you can disable the ref counting and you have a bunch of options for PODs. The real downside is that once you get other people onboard, they pretty much expect C#-style scripting. And the tooling is SO much better out of the box for C#. I'm still considering a small plugin feeding on some generated json or on the headers directly and deriving colouring and auto completion from it. For the Splitterwelten we had a simple code editor based on the Qt texteditor framework which at least could highlight and autocomplete all symbols registered at the AngelScript runtime. I never did this again, and I miss it.

Don't know what point I was going to make, sorry. Just dropping anecdotes, I guess. The support is great, though! That guy is still present and responding pretty quickly at the GameDev.net forums if you ever have questions.

[oahda]

Always interesting to hear about other experiences!

I’ve used AS a little bit before so I mostly know what to expect, but I did do a better job of researching it this time so now I know it even better, and it feels good~ For example I’ve made sure to cover things like the generic calling convention on platforms without support for the native one—luckily very quickly done by using the autowrapper add-on so I didn’t have to do much extra work manually!

Overall it seems like most of the things you mentioned regarding registration are fixed by the AS type traits function, but it’s only available if using C++11 or above so maybe you haven’t had access to it? Still have to rebuild to check any registration errors until they’re fixed, but takes away the guesswork for the most part. c:

And yeah, I try to keep memory management and GC on the part of AS to a minimum—reference types are generally just mapped C++ types managed by the parent application (entity components, for instance) that can’t be directly instantiated in AS, and value types are POD or RAII (maths vectors, strings, things like that). There are so many neat add-ons out of the box too that save me from a lot of pain!

And yeah, only me ATM!

Would’ve been really nice to have C++ (static) reflection to automate all this binding, but alas, that didn’t make it into C++20 either and I’m still stuck on 14 for now anyway...

[qMopey]

I don’t believe in automated binding