Saturday 21 February 2009

Best tagline for a roguelike ever

Gruesome

It is pitch black. You are likely to eat someone…

Something Awful Saturday featuring Goon Tower

From the guys who brought you Classic Video Game Covers, comes the Dwarf Fortress floor in Goon Tower:

Next week I'll be looking at the Goon City.

This ongoing series is dedicated to the FAAngband 0.3.6 release.

Thursday 19 February 2009

8PRL – the 8x8 pixel Roguelike Competition

The challenge: create a roguelike that has a total resolution of 8 pixels by 8 pixels, each pixel of which is an RGB LED display.

It's not entirely clear, but I suspect the RGB display implies a total colour depth of 2^3 = 8 colours (red, green and blue set to on or off state).

It may be possible to vary the intensities more: certainly the screen shot suggests this (orange is displayed, which doesn't fall into the above criteria). I’d suggest that you can have a maximum of 28 distinguishable colours in a 'screen based' RGB display, based on what Leon Marrick achieved in Sangband.

For consistency, it’s probably best to refer to them as follows:

D - Dark Gray

w – White

s – Gray

o - Orange

r – Red

g – Green

b – Blue

u - Brown

d – Black

W - Light Gray

v – Violet

y - Yellow

R - Light Red

G - Light Green

B - Light Blue

U - Light Brown

p – Purple

P – Light Purple

t – Teal

m – mud

Y – Light Yellow

i – Dark Pink

T – Light Teal

V – Light Violet

M – Mustard

I – Light Pink

z – Blue Slate

Z – Deep Light Blue


If you’re especially concerned, I can give you a copy of the RGB values that Sangband uses, or you can ask at the original 'challenge' location. Note that this is a maximum number of colours: you’re free to use less for bonus points and to make the game easier to understand.

Note the reference hardware supports a buzzer for a speaker and single life bar across the top consisting of a single colour status LEDs. It has a limited number of controls: a d-pad and two other buttons.

I’ll post my design separately to avoid influencing the discussion further.

May the best pixels win.

Wednesday 18 February 2009

More Borg Call Graphs

I'm experiencing this phenomena called being busy that I don't normally allow to affect my blogging or development work but appears to be doing so more as I age. I believe this is related to a phenomena called sleep that seems to be an requirement I wasn't previously aware of.

In lieu of anything intelligent to say with this update, I present more borg call graphs. Thanks again to darke for generating these. From his notes:

These ones are 500k turns, with a level change every 1000 turns to try and simulate someone actually playing it. Took about 20 hours to run through the tests with all the instrumentation turned on.

The first is the entire graph. The second one focuses on the playing "dungeon" side of the tree. The last one focuses on the dungeon generation "cave_gen" side of the tree.



Saturday 7 February 2009

Classic Video Game Covers

I'd been enjoying reading this thread on Something Awful all morning and just started getting to the roguelikes:


And earlier:



and


I hope the above is fair use - please see the 28+ page thread for the originals.

Monday 2 February 2009

Borg call graph

I would just like to publicly thank darke who has made an excessive number of bug fixes on the SVN version of Unangband in the last few weeks - including picking up some golden oldie bugs.

This is yet another example of why making your source code easily available is a great idea.

A number of these bugs were identified by using the so-called dumb borg, an automated Angband player developed by Leon Marrick. The dumb borg is different from other Angband borg implementations because it cheats, and in the Unangband version, is prone to jump rapidly from level to level.

Darke's just sent me a call graph for a 1000 turns of borg, which I'd like to share with you. The borg is heavily biased towards spending time in dungeon generation, but it's still interesting to see which code is the most heavily used. (Click on the picture for a higher resolution version).

Many of the remaining bugs were identified by darke porting the Angband source code to C++ (by renaming every .c file to .cpp and recompiling). C++ picks up a lot more warnings than the venerable gcc compiler I've been using - and most of the warnings have identified logic and syntactic errors in the code that he's gone on to fix.

Sunday 1 February 2009

Linux.conf.au slides

For those of you who attended, the slides to my presentation are available here.

For the rest of you, I hope you can figure out what I'm trying to say.