jsbarretto

jsbarretto, 11 days ago

I've fixed the horrible near plane clipping issues, added very basic collision resolution (literally 'snap to the top of the highest floor') and given him a drop shadow.

A video of a red triangle wandering around Peach's Castle, the starting location of Super Mario 64. The level geometry is flat-shaded and the resolution is extremely low, just 120x80 pixels. The red triangle has a small drop-shadow below it. The movement is jerk and awkward, and the collision resolution is half-baked: the triangle frequently snaps to the top of walls and overhangs when moving near them. Additionally, the frame-rate is low, no better than 20 frames per second.

jsbarretto, 10 days ago

I have implemented collision detection and resolution with the geometry mesh including wall and floor sliding...

...all without a single division operation (they're slow).

I spent ages trying to fix a mystery unsolid floor bug only to realise that the triangle frustum check was overflowing internally. Fixed-point maths sucks, makes me realise how painless and elegant floating point is in comparison.

jsbarretto, 10 days ago

Are there any 3D modelling gurus that feel like taking a stab at an ultra low-poly Mario? I'm abysmal with Blender. Here's my attempt at simplifying the existing model, down to 172 polygons. Ideally, I'd like to get below 100.

jsbarretto, 8 days ago

Texture mapping, mipmaps, and transparency support for today. Oh, and the game can now switch between stages.

I'm surprised by how performant these ended up being. Texture mapping roughly doubles the cost of the rasteriser inner loop but nothing else, mipmapping is basically free, and transparency support - after a lot of prodding - compiles down to a single assembly instruction in the inner loop.

That said, none of them are free so the game is gradually slowing down despite my best efforts.

A zoomed out view of Peach's castle with mipmapping enabled. Because of this, the textures in the scene are less noisy and the shape of the castle is clearer and easier on the eye (less visual noise).
A close up screenshot in front of the castle showing textured walls, a semi-transparent fence, and shaded textures.

jsbarretto, 8 days ago

Here's a video of it all in action (minus mipmapping, I got that working less than an hour ago): https://youtu.be/K0ye--qxt-w

jsbarretto, 20 days ago

@gsuberland Cherish them, in time you'll recognise them as the true refugees of the scalp that they are

jsbarretto, 26 days ago

@nixCraft The problem with "simple interfaces" like text is that you end up needing to implement unique parsing and error handling code in basically every tool to properly deal with the unstructured interface that it's forced to accept, and now suddenly each tool is not so simple any longer.

jsbarretto, 1 month ago

A few months ago I got an old Thinkpad up and running again to use on my commute. It's got a pretty tiny SSD on it so to save on disk space I decided I'd install a distro that leans into system-wide packages instead of containers.

I've been very careful to only install things from system repositories (with a few minor exceptions for things I've installed from source).

A few months later and things are going great. I spend much less time fucking around trying to make things work. It's blissful.

jsbarretto, 1 month ago

This little laptop is really a joy to work with now. It's incredibly stable, the DE and apps have a really small footprint, and all of the software I have installed just works.

jsbarretto, 1 month ago

On my main machine, on which I have several containerised apps, I've hit the following containerisation issues:

• Firefox demanding to be made the default browser every few weeks (it already is)

• Gamepad inputs not being passes to apps seemingly at random (not permissions-related)

• Some apps being unable to start a Vulkan context

• Dark/light theme not applying for some apps

• Never knowing where to find config files for any given app

• CSD occasionally not working for some apps

jsbarretto, 1 month ago

@RL_Dane It seems interesting at first glance, but honestly, I don't feel like I'm personally very interested in containerising my life. Maybe there are folks for whom it brings a lot of utility, but in my case I'm in the habit of just not installing software I don't trust: so the convenience of installing things system-wide will usually win every time. If I have something accessible on the network, then I'll usually just use a dedicated docker instance for that.

jsbarretto, 2 months ago

@0xSim that's a funny place to stash dru---

oh. oh no

jsbarretto, 2 months ago

@janriemer Nice! If you ever find the type signatures confusing, remember that you can use .boxed() pretty much anywhere to massively simplify things: https://docs.rs/chumsky/1.0.0-alpha.6/chumsky/guide/_00_getting_started/index.html#advice

Also, the examples in the docs are quite numerous and should be fairly helpful. The crate examples show you how to grow everything into a more non-trivial design.

jsbarretto, 2 months ago

@janriemer I would recommend against checking equality for errors. Chumsky doesn't have stable guarantees around exactly how it goes about constructing an error, so future versions may change the exact output a little. Usually, .has_errors() is the best thing to use in tests.

jsbarretto, 2 months ago

@janriemer Recursion couldn't be easier!

let a = recursive(|a| /* definition of a */);

Make sure you don't end up with accidental left recursion though. Check out the docs for the new pratt parser combinator if you want a really neat way to do complex expression parsing.

jsbarretto, 2 months ago

@janriemer So... Yes, memoized does exist, and does allow the parsing of left-recursive grammars. However, it does come with some overhead: if you can rewrite your grammar to be right-recursive, you'll almost certainly be happier in the long run (this goes for pretty much any top-down approach to parsing out there FWIW).

I'd be interested to know what the right-recursive section of your grammar looks like: the pratt parser can almost certainly handle it, whatever it is.

jsbarretto, 2 months ago

@janriemer Both of these cases can definitely be handled by a pratt parser! The first cases (&& and ||) are just regular binary expressions. The second case (method calls) is just a postfix unary expression, but where the 'operator' is .method(<expr>). Remember that 'operator' doesn't need to be just a single token, it just needs to be 'self-delimiting' (i.e: outside the precedence hierarchy)!

jsbarretto, 3 months ago

Also, the previous owner of the house has left exposed, live wires here? Under a BATH?! With the only thing standing between live and neutral being a poorly tied knot? Christ on a bike

jsbarretto, 3 months ago

(I am not very good at Inkscape, sorry. Ah well: amateurism makes it seem more local and relatable, I guess)