Thread with 14 posts
jump to expanded post
ISO C committee, a failed standard effort:
- dysfunctional, couldn't standardise a simple improvement if their lives depended on it
- arcane, incomprehensible specs that nobody reads
- all volunteers. literally nobody involved gets paid
- they have the gall to paywall it??
Khronos (Vulkan, OpenGL) and MIDI specs:
- long track record of pragmatism
- written by big companies that have hardware to ship and no patience for bullshit
- down-to-earth, sometimes refreshingly casual, easy-to-understand specs
- everyone involved has a real job
- free :)
these are broad generalisations based on vibes and some amount of first-, second- and third-hand experience, so i can't go into detail, sorry. also, please don't be insulted if you worked on the C standard, this is not a personal attack but a statement about the system as a whole
‪(i'm not here to talk shit for internet points, i think the state of the C standard is a very sad thing and i want to understand how others can avoid it)‬
,
@marisadoom@tech.lgbt
@hikari I'm a C programming veteran and I forgive you.
I would not call Vulkan easy-to-understand, nor hold OpenGL as a bastion of a spec written by "no-bullshit" ppl (the person who wrote Glium burnt out for a reason), but the other points are fair.
@cr1901 they are easy to understand compared to what the same specs would look like if the C standard committee had written them. the inherit complexity of the subject matter makes it quite hard
@hikari Yea, and I'll never not be bitter that OpenGL evolved from "draw these triangles for me and transform them via a matrix stack" to "buffer binding/shuffling", (GLSL is okay...)
@cr1901 having gotten deeply familiar with both versions of gl i unfortunately must object that the buffer binding/shuffling version is much better in practice and the old version's only real advantage was beginner-friendliness
@cr1901 …but friendliness at a superficial level. doing anything complex with the old system is a nightmare
@cr1901 the new system is easier to build an abstraction over; write the shuffling code once and you don't have to think about it again
Not A Serious/Constructive Response
@hikari If it's a nightmare to do anything complex in 1.x, then don't do anything complex :).
Most of the time, all I wanted was to draw triangles to a screen and maybe "decorate" them with a fragment shader (Phong lighting).
I've written a vertex shader > 10 lines exactly once (for a class).
Not A Serious/Constructive Response
@cr1901 check out the source code of my personal website :)
Not A Serious/Constructive Response
@hikari Will do once I get home.
Not A Serious/Constructive Response
@hikari Your website reminds me of learning fractals via fragment shader: