WorldSpawn official WIP thread

[Mugwump]

That's great! Now you got me pumped for your utility and I'll be eagerly waiting for the release.

You decided that meant game developers.

Heh, fair enough.

[MadGypsy]

I rewrote the entire old ui code. I know it looks like I'm just one baby step further than I was in my last post but technically I am substantially further. The real accomplishments are in my code. I'm just messing around with ideas. All of this stuff can be recustomized with ease to look entirely different.




I noticed these images have some artifacts and darkening. The actual has brighter reds on the buttons and a really smooth gradient for red to black transitions.

[Mugwump]

Did you just implement an option to make Worldspawn serve as a dynamic wallpaper?!

@darkening It's a common problem with YouTube videos. YouTubers usually have to crank up their brightness controls to make Quake videos look like they should. Maybe it's a similar issue?

[MadGypsy]

@dynamic wallpaper

no...lol. That's just my wallpaper. Bout time to change it too. That was way back when I was first getting particles to work. The images are just to illustrate a working application shell.

[MadGypsy]

Dynamic ScrollPane -

automatically adds vertical or horizontal scrollers based on "overflow"
follows resize scaling to the nth child
clickable scroll track


ScrollPane was instantiated with width 400 fixed / height rest.

[MadGypsy]

Yuge milestone. I just merged all 3 of my projects into one mega project. I don't mean that I just dumped all the folders into the same source either. Some things that seemed like they belong with one thing actually made sense to be with a different thing. An example is initial parsing of BSP. You would think that goes in the engine... and it definitely could but, my content manager knows everything about loading and parsing files. To me, it makes more sense for the parsing of a bsp to lumps (and miptex lump to final images) to happen in the content manager. There were more examples of this. I reorganized things with this type of stuff in mind. As is stands, I have a 3d engine, a content manager and an Object driven gui system living in harmony in one really big project. I really really hope that these things being together will stop me from losing track all the time from one project to the next. My next step is to create 3 separate "Main" classes, 1 for each thing. This way I can set whichever one I want as the document class depending on which thing I am working on. I can't really think of anything else I can do to make this as easy as possible for myself. It was always the plan that these things would all be parts of one thing it just took a really long time to get all of this stuff to a point where putting them together wouldn't be a mess. I am very satisfied with the current structure. It all makes sense and is very easy to search. Not to mention Ive written most of this stuff exactly 7129 times so, it's optimized, commented and documented to the nth degree. In many cases there isn't a single line that is not explained.

more...(while I finish up this coffee)

The object driven gui thing uses the content manager. The 3d engine uses the content manager. Both of these things actually have a dependency on the content manager. That's fine because, both of these things were intended (to some degree) as highly elaborate examples of the power of extending my content manager. The gui thing (hereby known as App Reflector) works on the 2d stage (always the topmost layer) and of course my engine works in a 3d context that is at least under the 2d stage, and at most, also under 3 more 3d contexts. That means that there is no way for this stuff to conflict. Worldspawn and App Reflector are completely oblivious of one another. They can't even share the same stage. It's gonna be bad-ass when I start bringing together examples of all this stuff working together.

But, that aint gonna even begin tonight. I want to learn how to build a worker pool and figure out how to use it properly. I want it to be a lot like the content manager... always ready to give results without any primer code. For years and years the best I could do with the content manager was only adding one listener and having the multiple necessary listeners handled internally. This meant in order to open/save a file I first had to make at least one listener, import the events for that listener, etc, etc. I got that shit down to NO listeners. You can call a content manager function out of the blue, anywhere with zero setup. You don't even have to import the content manager. I want my worker pool to be that simple.

var bytes:ByteArray = Content.getWad(mapname);

everything is this damn simple for I believe 27 file types...might be 23

oh not only is it that simple but, if you have a zip reference loaded it will even go right to it in a zip. You have a zip reference loaded but it's not in the zip....you cannot run, you cannot hide, my content manager will go as far as the security sandbox allows to find your file. You don't even need to supply a path.

[MadGypsy]

I just merged the stream player as well and my entire project comes in at 59 classes. That is not including a single interface (of which there are 3 and 1 abstract). A rough estimate made by rounding my biggest classes and multiplying the number of things that implement an interface by the amount of lines in any one of it's classes... I get about 26000 lines as my result. The real number is probably even higher, maybe by quite a bit.

That's a lot of code and much of this has a Haxe version as well. I don't program in haxe the as3 way (mostly) and due to compiler target exceptions my Haxe more than doubles the number (except there is no haxe streamer but that's only 8 classes and 2 of them are bullshit...20 lines or summin).

To sum it up, in about 10 months I've written something like 60,000 lines of code and I would be willing to bet the number is actually much higher but, I'm not counting stuff I didn't keep or became what I have now. I'm also not counting code I have posted, of which I generally type out when its the gist stuff.

[MadGypsy]

Well, I got Workers (ie multi-threading) to work but, IDK. I'm not all that impressed with the performance. In my image, the code is super duper simple. I have a local loop count to 10000 and simultaneously have a Worker count to 10000. All this gets kicked off by simply clicking the stage. Even though I let the worker start first it still finishes last by less than a second. That's shitty though. All it has to do is count to 10,000. How slow would it be if I had it doing something much more complicated? I need to run more tests. There is the possibility that the worker itself is not creating the delay. In other words, maybe the worker is just as fast as not using it but sending back the results message always has a 400ms delay... or something like that. All I know is, I'm gonna drop a bomb in the workers lap and simultaneously process that same bomb in the main thread...if there is some seconds-long difference workers are going in the garbage.

Honestly, after reading numerous posts, tutorials, etc regarding how flash handles workers... I feel like adobe really dropped the ball on this feature. It's not very user friendly and the way the thread is being instantiated is little more than a sloppy hack IMO. Basically I'm just running code in a separate, "on-the-fly", invisible swf...it's not impressive.

[MadGypsy]

Whatever the delay is with threads, I don't think it has jack shit to do with whatever work you are asking the thread to do. The numbers in the local and worker fields are the milliseconds that each count ran before it completed.


count to 10 thousand - diff 582 ms


count to 1 million - diff 565 ms


count to 10 million - diff 718ms

[Mugwump]

You don't even need to supply a path.

Good thinking. Too many programs are too stupid to even look for a file inside their own folder.

in about 10 months I've written something like 60,000 lines of code

That makes around 200 lines per day. Is it a good average for programmers?

[MadGypsy]

@That makes around 200 lines per day. Is it a good average for programmers?

I have no clue but, it's really around 200 KEPT lines per day. I do the same thing over and over but differently. The idea isn't even necessarily to finish said thing. The idea is to understand how everything could be better to a degree where I stop writing the shit that will need to be fixed later to begin with. I'm past languages, constructs, patterns and all that shit, now it's all about training myself to write the tightest stuff the first time. One day, even if it's just for 5 minutes before some prodigy kid takes it away from me, I am going to be the best programmer in the world. NO BARRIERS...NONE.

[MadGypsy]

I have officially made up my mind that I am no longer a flash programmer. I have mastered the language and outgrew it as well. From this point, all of my work will be done in Haxe and I intend to start completely over. I don't have to start over. I have perfectly good Haxe code already. I want to start over.

Content manager, object based ui, stream player, engine ... all of it. I'm throwing it all in the garbage and starting completely from scratch. This will be my last and final scratch start. I'm going to buy a huge chalkboard and do something I have never ever done in all my years of programming. I'm gonna write a flow chart... it's gonna be sick as fuck cause there is a whoooooole lot of stuff to graph out.

[Mugwump]

Damn! Are you scrapping all of this just because of that workers issue? I was waiting for your content manager release to try its submipping capabilities with my custom textures. Oh well... I hope the next version won't take 10 more months to make.

[MadGypsy]

pbbbt I'll knock it all out in a saturday. At this point I could practically recite the entire program from memory. I didn't know you wanted a wad thingy. I'll bang one out for you. MAYBE tonight...all the groundwork is laid, I just need to decide how to display it.

@wirker being my reason

no, not at all. I still love flash, it's a great language for some kind of middle ground between c and javascript and I could see myself using as3 to quicly solve a problem that I could port back to haxe. Ive been considering dumping flash for 10 months...lol. I'm really just sick of the lack of typedefs. Haxe turned me on to typedefs and now I can't live without them.

[Mugwump]

I didn't know you wanted a wad thingy.

Umm, I explained what I was looking for and why about a week ago in post #268 and told you I was waiting for your content manager release 3 posts later...
Edit: Oh BTW, I incorrectly mentioned dithering when talking about how Wally/TexMex generated the submips. It would seem that I had the wrong idea of what dithering actually is. I meant blending.

I'll bang one out for you. MAYBE tonight...

Heh! Firstly thank you and secondly, I'm not in such a rush. I can wait a while. It just seemed that you were almost done with it and that the release would've been rather soonish, so I was slightly taken aback when you decided to scrap the whole thing. But no biggie.