trfd @trfd_

Co-founder and graphics programmer @endroad_team. Ex developer @Amplitude endroad.fr Nantes, France Joined April 2011

Tweets

280
Followers

542
Following

381
Likes

2K

trfd @trfd_

2 years ago

@olson_dan @sirGustav It would make sense in my head as a syntactic sugar for "locally defined sum type". But not sure if this what proposed.

0 0 0 51 0

View Details

trfd @trfd_

2 years ago

@0x21376B00 @PierreTerdiman A statically typed python with braces and templates (and c perfs) would be my dream language tbh 😀

0 0 0 100 0

View Details

@SebAaltonen It's really hard not to feel sad at the state of the world when we see that. But mathematicians already have a solution for that number based on the cardinality of sets: 0={}, 1={{}} 2={{},{{}}} ... If we had to give up on a normal numerals, we might as well pick this one

0 0 0 394 0

View Details

trfd @trfd_

2 years ago

@olson_dan That's smart! Didn't think of this one. Pretty sure some CTOs would have pushed to move to Rust if possible (safely) for consoles, even against team and techno inertias etc., just bc the time/money saved using Rust in long run is huge and ppl would not be as dismissive (I guess?)

0 0 0 28 0

View Details

trfd @trfd_

2 years ago

@olson_dan Instead I have been thinking about an alternative that transpile to C but very hard to convince anybody to work on (or fund) this. From business pov it's not worth the investment, harder to onboard and support teams, reduces possiblity of outsourcing or uses of third parties etc.

1 0 1 23 0

View Details

trfd @trfd_

2 years ago

@olson_dan In the end there might be just a handful of people that are both qualified, willing and have resources to work on something like that. In our case we would have adopted Rust (even with it flaws) for our new engine if we were able to ship on console with it.

1 0 0 32 0

View Details

trfd @trfd_

2 years ago

@olson_dan Imho it's also a limit for us (game devs) as most of the effort of making c++ safer goes through the std and not the compiler. IDK if Rust is the best solution possible for safety but having to rely heavily on one impl. doesn't sound really appealing (to me)

1 0 0 19 0

View Details

trfd @trfd_

2 years ago

@Donzanoid Ah yes indeed ! That would be a decent compromise. Gonna try something now! I've also been considering moving the ecosystem to clang exclusive support just to use this gcc.gnu.org/onlinedocs/gcc… (and also benefit fully from the clang tooling)

0 0 0 57 0

View Details

trfd @trfd_

2 years ago

@ChristerEricson Thanks! That's funny you would say that, I recently removed our generic assert macro and replaced it with `EXPECT`, `SAFETY_CHECK` ... along w/ usage guidelines to help the team choose between our rust-like result and the right assert. But still does not feel clear-cut enough

0 0 1 94 0

View Details

trfd @trfd_

2 years ago

@Donzanoid Our pipelines are entirely based around Bazel too! I guess that's why we entertained the idea for so long. More worried about editor support (syntax highlight, language server ...) tho. Historically it's the stuff that always gets me. Much harder than language itself imho.

0 0 0 115 0

View Details

trfd @trfd_

2 years ago

@Donzanoid Thanks! I'm going to revisit macros now ! I'm very tempted to write a parser or compiler plugin for that and also other stuff like rust's enum or more powerful macro system. Just afraid of tooling support which is always a lot of work for production ready languages

0 0 0 110 0

View Details

trfd @trfd_

2 years ago

@Donzanoid Totally agree! I introduced them in our codebase recently and it's so much better to work with C++. However I'm not a fan of my implementation (esp. the "try"). Still not as nice as Rust's. Do you have any recommendations for a clean and simple API?

0 0 0 200 0

View Details

trfd @trfd_

2 years ago

@kenpex Good points! If we admit normals are just a model for something more complex and consider normals together with how they are used in a rendering eq (e.g. L.N), we should get this "scaling" with density of the PM toward a classic solid diffuse model (analytically) I guess ?

0 0 0 73 0

View Details

fabs(); @raspofabs

2 years ago

Ever wondered why Design Patterns didn't take off? How did they become both all-pervasive and soulless? Those were the sparking thoughts behind this book. It's taken four years of research and way too much of my spare time. Go check out Unresolved Forces: unresolvedforces.com

0 8 39 4K 23

View Details

trfd @trfd_

3 years ago

@maxberger That's madness! Who would be left to rule us ?

0 0 0 7 0

View Details

trfd @trfd_

5 years ago

@Atrix256 @kareem_carr @jp_axs4ll Will you coin this method as Monte Carlo Knowledge Acquisition ?

0 0 1 0 0

View Details

trfd @trfd_

5 years ago

@olson_dan By far console support is our main concern (Doesn't seem to improve either which is even more concerning). But we have the chance of being a relatively young studio which make everything easier. (And we already started adoption of rust for everything tool related)