What Can We Do

[Lincent]

Too bad its breaks broke, now its a wreck.


I have a better idea, remake Ace of Spades OpenGL, with an improved Openspades engine!

[Kuunikal]

Too bad its breaks broke, now its a wreck.


I have a better idea, remake Ace of Spades OpenGL, with an improved Openspades engine!

How about you stop being so pessimistic about the project and actually try to contribute since you seem to be so active on this topic?

[Chameleon]

Email @ PM?
I'm not doing anything with that document right now... We use titanpad, but we could go with google docs

[Lincent]

How about you stop being so pessimistic about the project and actually try to contribute since you seem to be so active on this topic?

Because this "project" is going to be a complete failure.

[longbyte1]

hey now that the admins are back Lincent, you ever feel like you need some disciplinary action?

[tin]

It's awesome that you have the will to keep this game going, so if you have any ideas/suggestions regarding future improvements it would be nice to post them in the anniversary thread so we could gather various opinions at one place.

[Zekamalikyd]

i heard there were people talking shit last month

[Lincent]

hey now that the admins are back Lincent, you ever feel like you need some disciplinary action?

Oh no, not that.
Hey, now that you've decided that you are the president, do you ever feel like people need to go to jail?

i heard there were people talking shit last month

k

[bloodfox]

Tin, put some sense in this douche, please.

[Lincent]

Tin, put some sense in this douche, please.

You're just mad, because I made a joke about leaving, and people thought you were the one leaving.

[tin]

Take your personal discussions and issues to PM.

[Marisa Kirisame]

And now we can return to this damn topic. This one is almost 7 pages long.

1. *i use bad words*
2. im from the 1.x forums

To be blunt I think a lot of the people on the 1.x forums "used bad words". It's par for the course. Also, what was your name on there?

why does nobody care about this game anymore
what is the point of moping around and complaining instead of taking action
why dont we try to contact ben

Because it's been tried and tried and tried. Anyway, enough of the depressing shit. Wait, there's one more thing I need to point out:

Unless you are Ken Silverman, you will not get a sufficiently fast software renderer working.

I tried it with Iceball and still didn't get it working at a good speed. yvt also tried it and didn't get much speed either. It's just not practical.

Do the programmers in your community a favour, and get a better computer.

Now back to the topic.

This forum only has one purpose now, and it is to develop its successor.

Oh shit this forum is now Homestuck. (NP: Elevatorstuck)

Anyway, you wanna do something? Okay, let's start with design.

I say start with code. But hey, you've had a month to design things. If you haven't written any code yet, write some fucking code.

Second question is, are we going to start off with 0.75 compatibility or are we going to redesign the network protocol from scratch?

Redesign, but ofc we should look at AoS protocol's both good solutions and bugs.

The "world update packet" does help to keep things small. Unless you're whoever the hell was working for Jagex at the time and managed to bloat the shit out of the packet with fields which hardly ever change. In terms of network usage, 0.x < IB < 1.x - smaller is better. IB doesn't have the "world update packet".

AoS and IB both use ENet for networking. IB originally used TCP but that required more code than ENet did, and ends up doubling the latency even with TCP_NODELAY enabled.

If you want an example of how to do networking cleverly though, check out the Quake 3 network protocol some time, especially if you're going with C or C++.

Something to not borrow from 0.x: "I hit this player in the head at some point in time, trust me, I'm not going to tell you how but I did" - ask UnrealIncident about how well that works.
Something to borrow from 1.x: Packets are timestamped.
Something not to borrow from 1.x: Packets are "encrypted". Badly. And this "encryption" actually broke things on certain Wine setups.

Dynamic light should be done a lot more cheaply - light up not parts of models/blocks separately, but whole models/blocks instead.

Depends on your algorithm. If it's just a point-light diffuse+specular algorithm, that would have no real performance impact.

hipster languages like lua or js

Correction: JS is hipster. Lua is ghetto.

Instead of saying it's blocky, you should say it's "retro style", which will actually attract people due to this retro style hype.

I'd almost say "low-poly" but as anyone who's tried doing this in OpenGL would know that's a big fat lie.

For reference, the PS1 can fairly safely do 1,500 polys (tri or quad) per frame for 30fps. You'd want to halve that for 60fps.

Assuming we have a 16x16 chunk where 3 quads per column are draw, that's 768 off the bat, and we're definitely doing more than 16x16 at a time.

So unfortunately, we can't ride the Next Big Thing.

But we can basically shove voxels up their arseholes everywhere.

I never really understood how it's possible that smoke lags - alpha blits are cheap af for GPUs.

If you have a metric fuckton of them it's slow.

If you have one smoke particle per VBO it's also slow.

Why not we just go for something along the lines of C++? Easier for all of us to convert to and it's more up to date than of C.

You did not just call C++ more up-to-date that C.

With talk about Data Oriented Design being important to have and Object Oriented Programming being important to NOT have in gaming, I'd argue that C is the more "up-to-date" of the two.

Oh, and my eye's not a camera, so forget lens flare.

Eh, it's pretty cheap to add, you just use occlusion queries.

Then again, I'd say that bleach is pretty cheap to add to your diet.

a) How hard is Lua for Free-Pascal/C++/C# newbie?
b) Should asset checking on connection be hardcoded into .exe or can it be done in Lua? (so I don't have to dl whole game...)
c) How do I disable the toon effect?
d) How much stuff is actually hardcoded into .exe?
e) Why GM abandoned SW? It had LOD, which made it possible to run on 2.4GHz single-core. (I have i7 now, but performance matters)

a) If you've programmed before it's not terribly hard.

b) It should be cached and checksummed (using SHA2, not weakshit like CRC32... and MD5 is also weakshit these days). Iceball doesn't do it because I never got around to doing it, but that's what the dlcache/ directory was meant to be used for. Either way, if you can avoid roundtripping for every single file, it helps.

Also, if you want to you can actually make a mod and/or game that does the checksumming for you and dumps the files into clsave/vol/.

c) clsave/pub/user.json: "postproc": "simple",

d) Gameplaywise? Nothing. Unless there's a few helper functions I either forgot about or don't know about. But basically you've got interfaces to an engine that could let you make pretty much anything... but was still definitely designed for an AoS clone.

e) Because it was still slower than OpenGL on my Intel HD 3000 laptop, and not really any improvement on my GM45 laptop. Oh, and there's an experimental GLSL raytracer in the repo... which actually reached 60fps in some extreme cases on the GM45. No shit, raytracing voxels on the GPU (GMA 4500MHD by the way!) was actually faster than rasterising on that thing.

anyone feel like registering #aosofthefuture with Q?

Fuck QuakeNet. It's shit.

Seriously, just do it on the Fraction IRC (same IRC network that #iceball is on). As long as you don't start spilling shit over into other channels you should be fine. You'll also have a few people who know how things work but really really cannot be fucked with 0.75, but they'll still be able to help.

I also have a Mumble server you guys can use if you want.

Now all we need is BR_ on the boat and we'll be off!

And what is BR_'s area of competence?

Coded the first non-AoS game for Iceball: 3D Snake.

Aside from that, also made an example SMG mod that wasn't an overpowered sack of shit, and probably a lot of things that don't make him sound like a waste of space. Basically, he did some useful shit for Iceball.

Oh, and wrote some Pyspades scripts. I don't remember exactly what, but I believe there was at least one game mode in there.

Uhm, I'm also studying Python right now (not sure if theres a place for Python in your game). Maybe I can help in debugging or anything simple.

If you're taking the Iceball approach (send sandboxed code to the client), Python is quite possibly the worst idea ever. That shit is literally impossible to sandbox. Otherwise, it's not a completely terrible idea. The unfortunate thing is that it does limit your opportunities, though.

But hey, as long as the game keeps getting worked on and manages to stay moddable and hopefully reasonably safe from someone shoving a huge-arse virus in it, it should be OK.

I will say however that:
1. you'll need a native language, e.g. C, to get any decent performance, and
2. the Python C API is a royal pain in the arse to use.

Either way I suspect the project is using D. Mostly because I know that LeCom actually codes stuff, and he's said that he's started writing something in D.

----

Guys, where's your GitHub repo? Chuck your code on there and I might be able to help you with a few things.

----

EDIT:

Do you take a guy thats a source of a lot of gameplay ideas?

Everyone has ideas. You'll want something else to do really. Like code, for instance.

Failing that, this game could definitely do with someone who can draw moe art.

[backwardsbash]

And now we can return to this damn topic. This one is almost 7 pages long.

1. *i use bad words*
2. im from the 1.x forums

To be blunt I think a lot of the people on the 1.x forums "used bad words". It's par for the course. Also, what was your name on there?

Admin edited it. Used to say "fuck off lincent"

[LeCom]

Ok, a few things I don't agree with GM on:

It is very well possible to write a voxel renderer which is as good as or even better than Voxlap. It's hard, but not impossible. Voxlap is VERY deprecated code; optimization "rules" are completely different today (especially GCC with -Ofast vs manual ASM, and I do have reasons to claim that GCC is better than Ken). "get better PCs" is never a good argument, seriously. Look at Ken.

Then, the only disadvantage Python has from my POV is its performance. I personally get along very well with the C API (with the exception of classes), and it's better than Cython. I'm only worried that some bottlenecks can't be moved to C properly. The client is written in D just as you supposed. Since scripts are only server-side, sandboxing might only be needed for the server, but it's not very important.

The protocol I designed to transfer as few data as possible, while maintaining maximum flexibility. Atm, it should send slightly less data than 0.76 while the game is running. With some features, it might take up more capacity tho. I had thoughts of doing the timestamp thing via net latency average, to keep the data flow not too high.

Source is on github.com/lecomm/aof-client and github.com/lecomm/aof-server

You're saying the server shouldn't just accept hit packets and check them first; well, CAN the server actually detect fake hit packets, or are you talking of AoS's hack detection scripts

[longbyte1]

Well, now we have another coder on the boat. I'll tell you right now that I am definitely not working on the client, at least not now.