I wanted to resurrect my research notes because I have stuff to put up, but they had gotten a bit big and unwieldy. I didn't want to install a big Wordpress or other plug-in solution OTT thing because yuck, so I made my own with a bit of PHP. I wanted to just have a folder, where if I dropped a file of text in it would auto-magically appear in a list of links on the left-hand side of this page. Then you click on them, but instead of taking you to a new page it just grabs the contents of the file and smacks them onto another part of the page. They are just alphabetically ordered by file name, so I name them with the date in European format; YYYY-MM-DD. This also means that I can find the latest one easily to use as the default post to display. Super simple. Anyway, it worked. I used this PHP code:
Another thing that I did recently was complete a technical report documenting my experiments with different vertex buffer layouts in OpenGL. There exist only vague notions that interleaving mesh data memory should draw faster than other layouts, so I put it to the test, and recorded about a week's worth of drawing time on a couple of different machines. I wrote it up in a LaTeX PDF: buffers.pdf. Note sure if/where to publish it yet.
I managed to measure the time that GL took on the CPU with glDrawArrays(), and also used a neat tool called APItrace to gather the corresponding GPU times taken on two different commodity graphics processors. I see afterwards that there's actually a GL extension to query to GPU time taken by a function, but API trace is nice in that it gives you some visualisations too, which can show how much one is lagging the other. The short story was that the different memory layouts alternatives don't make any significant difference to performance.
I wanted to put my WebGL skills to the test so made an entry for the Ludum Dare 48-hour game making competition. It's called Dolphin Rescue. The entry page is here , and you can play it on-line at: http://antongerdelan.net/dolphin_rescue/