Steffe.se

I recently started on a small 2D physics engine project that I for now call p2.js. It contains just the basics, spheres, particles, planes. They can interact either through frictionless contacts or spring forces.

There are a few reasons why I wanted to roll my own engine, however, the most important one for now was the question about using typed arrays or not. Now I’ve got the answer to that.
I followed Brandon Jones simple instructions on how to use the typed arrays correctly, and as one of his slides implies, I was indeed barfing rainbows when seeing the results. Okay, maybe not barfing rainbows, but I must agree that I was a bit surprised.

In total I get a performance gain of about 30% when using Float32Array instead of vector objects such as {x:2,y:1}. This number, 30%, is just a very rough estimate, because it depends on a lot of different parameters. However, my implementation made it relatively easy to switch between these. Another thing that I noticed was that switching between ordinary Arrays and Float32Array didn’t affect performance much at all, though there are a few other advantages of using the latter.

The key to using Float32Arrays is to avoid creating new ones. In the physics engine case this can get a bit tricky since there are things added and removed to the simulation in every timestep. A good example is the contacts. When two geometries collide I need a new ConactEquation instance in the engine. This is basically a holder for a number of vectors, so making a new instance every time it is needed is a no go. To solve this I made sure these objects are reused in between every timestep, and if there are excess objects, I store them for later use.

The scene you see in the image above is a simulation of 900 circles trapped in container consisting of 3 planes. The number of solver iterations is 10. I can get reasonable results by using fewer iterations too. Rendering is made in a small 2D demo renderer I built with Three.js.

I’m going to make a demo page for the engine soon, though for now only the code is available.

As I had recently implemented some graph traversing code, I was keen on using it in Cannon.js. You may think, why use graph algorithms in a physics engine? The idea is a bit tricky but I’ll try to explain.

In the usual contact solving case, we iterate over all contacts in the system. For each iteration, we transfer impulses from one body to another. In the example case of stacked boxes, iterating like this will make the top box “feel” impulses from the bottom box as the impulse travel through the stack (this depends on how many times we iterate over the system and what iteration order we use, though that’s another problem).

Many solvers have a “tolerance” parameter and it is used to check when to stop iterating. If the total error (read: contact overlap) is small, then the solution is good enough and we can stop. The tolerance is compared to the *sum* of all errors in the system.

Say we have one stack of boxes on a static plane and a sphere on the same plane. The total error will include both the errors from the stack and from the sphere. We will stop iterating over all the contacts when the total error is below the tolerance limit. Say we reach M iterations. This means we compute stuff M times on each of the N contacts in the system (a total of N*M computations).

Now what if we split the system into two independent systems, one for the stack+plane and one for the sphere+plane, and run the solver once for each of the two systems? The stack will probably still need M iterations, but the interesting thing is that the sphere will only need one. Why? Because the case of a single contact does not need to propagate impulses, and it can directly report the exact solution.

So, in the big system we need M*N computations and in the split system we need M*(N-1). That’s great! And this strategy works for many other systems too. In most cases, we can get away cheaper by using a split solver.

But what about the graph algorithm, you may say. It is used to find the independent sets in the system. Yes, it will add some complexity. However, that computation needed is not as hard as the solve part, and it has linear complexity (with respect to the number of contacts and bodies). The solving complexity depends on the number of contacts times the number of iterations. The number of iterations should be linearly dependent on the number of contacts to be able to propagate all impulses across the system, and so we end up with a quadratic solving complexity.

There is another advantage with the split solver: the subsystems can be solved in parallel. But that is another story!

The CANNON.SplitSolver class is available in the cannon.js/dev branch, and here is a live demo where you can toggle split for a scene.

Recently I’ve been working a lot on the Node.js version of friendship-bracelets.net. Here’s a quick status report.

Code reduction

I’ve reduced the code to less than half its size by abstracting key parts of the code and compressing a few static JS files. I really love JavaScript – abstracting code has never been easier (and dirtier).
One interesting abstraction I made was an “edit resource” page. For each resource I have a Schema class instance that can help create an HTML form and then validate input from the client.

Nginx

Other news is that I’ve started using Nginx to serve static files and proxy to Node. It was really easy to set up so I will probably continue using Nginx. The only drawback is that I really want PHPMyAdmin for administrating my database, so I still have to run an instance of Apache… Perhaps I’ll find a solution to this later on. I will post instructions about my setup when it’s stable.

Chat using Server-Sent Events

Another thing I’ve done is a chat client. I was really excited while doing this because it is a whole new concept to the site. This way we can get even closer interaction with the users.
The tech behind the real-time chat is server-sent events, or more specifically, EventEmitter in HTML5. Earlier I was determined to use WebSocket, but since server-sent events is more well supported (on both client and server) and good enough for the purpose, I went that way.

Caching of Express views

I’ve probably mentioned how to do this earlier, but now I’ve tried it. When starting the Express app with the environment variable NODE_ENV set to “production”, I simply run
app.configure("production",function(){
  app.enable("view cache");
}); This makes Express cache the templates inside the app, and it makes the app faster. By also relieving the app from serving static files (using Nginx), this makes the web app perform really well. It almost feels like running the app locally when it in reality it’s on a virtual machine in a datacenter somewhere else.

Mobile app thoughts

I’ve already started using jQuery mobile for the mobile site, but it will probably only make it more difficult for me to maintain the site. Using that will need different HTML for the layout, which duplicates that amount of code. I’m starting to think maybe it’s better to just add some CSS when on mobile instead.

Express: multi-language site

A question from a user came in, and he asked if the site could be translated into russian (he even offered help). Making an Express app support multiple languages is easy using e.g. i18n-node. However, making friendship-bracelets.net in multiple languages is probably not that easy. As I see it, we have 3 options for multi-language implementation.

1. Just translating the menu buttons and some of the text. This will encourage people comment in their own language (if the site is in your language, you will probably write text in your own language). I think it would be really confusing if everyone posted stuff in mixed languages.
2. Separating the languages into own sites with separate content. One site will have its own set of content and the other won’t have it. Not really cool and not really a good option in this case.
3. Let’s say we have two sites of different language, A and B, and they share all content. Should all content in A also be visible for users in B? For some content, yes. What if you post content in your own language site A (english) and get a comment on it in site B (russian)? It gets more complicated than this when you think about it, and there will be special rules for just about everything.

Discussing this with the mods lead to a decision to do nothing about the multi-language question. However, It would be really cool to try out i18n in the future (I never tried it before).

I’m getting closer to the core of friendship-bracelets.net… Pattern rendering. It is the most vital part. This is what almost all users go to the site for. I’m going to make it as good as I can.

One advantage with doing this in Node.js and JavaScript is that if there is a need for client side rendering, the same code could be used. An example could be to change the colors in the pattern interactively, to check whether a set of colors are good looking together or not.

Many things are falling into place now. I want to launch a beta testing site soon but before that I need generators and server-side rendering for all types of patterns… And probably a lot of other stuff that has nothing to do with patterns at all. I hope you can wait.

The comment boxes on friendship-bracelets.net are like small web browsers. They load stuff dynamically from other places and view only what you need. You can post your comment and a script will save it on the server for you.

The code from the old version needed to be rewritten for various reasons. The first is that the code was a monster… Second is that it did not load things comments from the server on the desired format.

Need to get some sleep now… Over and out.

The generators are the most important things on Friendship-Bracelets.net. Users use them to make their own pattern, either of type Normal, Alpha, Kumihimo or simple name patterns.

There has been a lot of problems with the generators since the site started. The first normal generator was written in pure PHP, and the users had to interact with it by writing “pattern code”. This code looked somewhat like this:

DDDD
ddd
RRRR
rrr

Each letter in this code was representing a knot, and each row represented a row in the pattern. D meant “forward knot”, d was backward knot, R backward-forward knot, and r forward-backward knot. Can you imagine writing a pattern this way?

Another drawback with a PHP generator was that the server had to do all the rendering and send back the results to the user every time he/she updated the code. This was very heavy for the server.

Later on, I made the generator in Flash. I really hated Flash (I still do) but it was the only reasonable way to make an interactive generator at that time.

By that time the site had the same generator code written in both Flash and PHP, which doubled the code maintenance time for me. And I had no experience in unit testing, which made everything a whole mess.

Now everything is going to be written in JavaScript, and all the code is going to be unit tested. I’m really excited that the site finally is getting simpler to maintain.

Both search and newsfeed are two things on the site that I struggled with the most when I made the first version of the site. Can’t see any problems with doing that today, but I guess that’s what they call experience.

I added support for guestbook and comment search upon request from the users. For the newsfeed I made sure there are no dead links that the old version had a lot of.

There has been ideas on making merged newsfeed entries too, such as “User1 and User2 commented on your pattern” instead of two separate entries. I guess that will have to be a future project. There are a lot of basic things that I want done first.

Building a native app is cool, though a bit unnecessary in my case. For Friendship-Bracelets.net there is already a website with already tested functionality and interface. Luckily, there is a surprisingly simple way of making the site mobile friendly. This is using jQuery Mobile.

The app you see above only took about 10 minutes to make given my site base code. It can do most things that my site already can, since jQuery Mobile adapts to most existing things in the HTML. Of course, it will need some polishing.

Worked some more on the new site, this time I got the signing up process done, including the email sending part. Or at least the first version of it.

I also added some nice information that the users will see if the database disconnects. Probably lot better than a simple error message like “too many connections”.

Since last time I’ve also signed up for my first Cloud service. I will run the new site on that, no doubt. What it means for you as a user is not much other than a stronger guarantee that the site is always up and running. The Cloud servers will run even if there’s a power outage since they have diesel generators backing them up.

Since I now got the cloud and a basic draft of the site, I hope to release an alpha version of it soon enough. There is just a few basic things I need to get working before this can happen, such as being able to change password.

Friendship-bracelets.net has been up for a long time without updates, and it is now getting a bit old. It’s time for me to rewrite it all in HTML5!

Why rebuild it all?

Rebuilding it will take a lot of time, but it will be worth it. The site will…

• become faster
• support more cool things
• become more secure
• be easier for me to maintain and develop
• have more nice looking URLs
• have new HTML5 generators with more features than the current ones
• have a developer & moderator blog
• feature better photo uploading
• probably be supported on more devices
• …and more!

Obviously, everything should be at least as good as it was before, most things will hopefully be a little bit better.

How far I got

I guess what you guys want is a screenshot, so here you go!

It is hard to explain to non-programmers exactly how much has been built of the site, but what I can say is that I’ve got the basics done.. Such as:

• Page layout
• Login and out
• Very basic HTML5 normal generator
• Most things regarding the guestbook, links and FAQ
• Static pages such as the Donate page
• Profile picture uploader
• Database layout and tools for transfer the old data to it

What is left to do is the rest of the site (duh!). I think that most of the remaining work is going to go more smoothly, as I’ve already learned how to use the tech behind the stuff above.

I’ll get back with more updates, stay tuned!