pyglet: full steam
Now that my thesis is handed in [still a presentation to go, later this week],
and exams are marked [still some assignments to mark] I can catch up to Richard's
work on pyglet. Lots more OS X support added, and a big refactoring of the
window creation process, which was ugly, confusing and impossible before; now
it's just confusing. Managed to do the big changes without breaking Xlib, but I
deliberately left Win32 to flail helplessly. I'll get to fixing that all up
after the next big change: bringing some consistency to the event names.
Public demos of Wii tomorrow and Tuesday at Melbourne Central, I'll probably go
and have a squizz.
Thesis has all the words and graphs and such in it now, they're probably just
all in the wrong order. Need to find some unsuspecting soul to proof read it for
Back into the thesis mindset now. Had a bit of a crisis yesterday when I
noticed that my entire argument was invalid. Then I went away, played some
Katamari, came back, and realised I was being ridiculous. All the math works,
and this time I wrote it down so I won't go crazy like that again. Few! That
should relieve some hypertension!
If the paper turns out to be on the short side, then I'm halfway finished at the
current page count. If it finishes at a more respectable length, then I'm about
30% through. If I stop now I'll almost certainly fail :-)
At least I can claim to have the best sources--what other computer science paper
can cite a Brothers Grimm fairy tale in its references? (The Frog King, if
Forget everything from the last post about the Neuros; the way to go is with a
Mac Mini. In the time between now and when I can afford one, I am using my 5
year old iBook with the screen off, which is silent, full resolution and
infinitely hackable. Now that Battlestar Galactica and House are both into
season 3 my TV fix is satisfied for the time being.
Back to the thesis
Approximately three weeks until it's due. I'm very, very, way, way, very
behind, having taken a bit of a holiday the last few weeks. The last couple of
days I revisited code I wrote near the start of the year so I can finalise and
run the experiments. None of it worked, so rather than actually *writing* my
thesis I'm still hacking code.
This is probably good from a thesis point of view: my new house has enough
separation between the computer and T.V. to make watching videos thoroughly
impractical. I tried a UHF transmitter/receiver, but it's extremely noisy and
the IR repeater doesn't work reliably at all. Not too keen on buying another PC
for the lounge room for environmental/sound/expense reasons, so I've been looking
into set-top network devices.
Top candidate at the moment is the Neuros OSD, which I suspect has very little
functionality, but at least is extremely hackable and will make a good holiday
project. The other boxes available (D-Link, Hauppage, LinkSys, etc.) have pretty
ordinary reviews, being not at all extensible, supporting limited file formats
and having cruddy interfaces. The Neuros doesn't do high-def or even component
video (curiously it has S-Video in but only composite out), but I can't see
anything that does and will do what I want (or be hackable).
Saw a possum and its baby in a tree outside my window the other night! Took a
couple of photos which didn't turn out at all (possum eyes act as excellent light
reflectors, for future reference).
ADSL reconnected and all's happy again. Well, except the PCI wireless card I
bought isn't playing nicely yet. It's one of those D-Link lotteries where the
chipset is not distinguishable by looking at the box alone. I got the unlucky
number, a Ralink rt61.
The good news is, Ralink have a GPL driver for it on their website. The bad
news is, it doesn't compile on AMD64. The good news is, there's an open-source
hackery of it which does compile. It works, but hangs the kernel after some
random time over [1,20] minutes.
For the moment there's a big CAT-5 cable snaking through the whole house--not my