3 ms·

Not my field but I don't understand the rationale behind choosing Rust. Given the performance and security concerns wouldn't Ada be the natural choice here? It

Not my field but I don't understand the rationale behind choosing Rust. Given the performance and security concerns wouldn't Ada be the natural choice here? It may not be the most fashionable language but it seems I read about security problems with embedded and IoT devices on a daily basis. Why not choose a language designed for this problem?

vertex-four 10y ago
Ada isn't actually particularly safe in any reasonable sense - or rather, it provides enough escape hatches with little obvious marking that it's costly (in terms of time spent in code review) to ensure safety. Safety of existing Ada programs is primarily a result of the engineering practices that went into them, more than the language itself.
- nickpsecurity 10y ago
  Nonsense. The designer looked at almost every spot where coding mistakes were leading to crashes or corruption. Then, made those safe-by-default wherever possible. They're described in modern form in the book below by Barnes. The effect was confirmed when defects dropped to around half what Mitre et al had been observing on C projects. There was only one case study I saw where Ada made no difference whose control group was C++. An anomaly that might have been same system reimplemented by same people, easily-correct system, or just both super-talented. So, it is a safe language with only exception being temporal safety w/out a GC. Rust delivers on that. http://www.adacore.com/uploads_gems/Ada_Safe_and_Secure_Booklet.pdf http://www.adacore.com/uploads_gems/Ada_Safe_and_Secure_Book... Also, SPARK Ada can straight-up prove common errors don't exist in that code with an automated prover. Most engineers would spend quite a bit of time doing that with pencil-and-paper examination or human-driven provers: https://en.wikipedia.org/wiki/SPARK_(programming_language) https://en.wikipedia.org/wiki/SPARK_(programming_language)
nickpsecurity 10y ago
It would be. Especially with mature tooling available from AdaCore. Seem my reply to vertex-four for links containing proof. Here's some examples in industry & CompSci just for kicks: http://www.adacore.com/customers http://www.adacore.com/customers Note that the authors are just doing research on seeing how well Rust handles the problem domain. There's nothing wrong with them avoiding Ada to explore another tool or problem area. They should consider each carefully if they were trying to pick best tool for real work.
naasking 10y ago
> Given the performance and security concerns wouldn't Ada be the natural choice here? Ada would be a good choice, but it's way more awkward to program in. It definitely has some nice features of its own that Rust should steal asap though.

vertex-four 10y ago

Ada isn't actually particularly safe in any reasonable sense - or rather, it provides enough escape hatches with little obvious marking that it's costly (in terms of time spent in code review) to ensure safety. Safety of existing Ada programs is primarily a result of the engineering practices that went into them, more than the language itself.

nickpsecurity 10y ago
Nonsense. The designer looked at almost every spot where coding mistakes were leading to crashes or corruption. Then, made those safe-by-default wherever possible. They're described in modern form in the book below by Barnes. The effect was confirmed when defects dropped to around half what Mitre et al had been observing on C projects. There was only one case study I saw where Ada made no difference whose control group was C++. An anomaly that might have been same system reimplemented by same people, easily-correct system, or just both super-talented. So, it is a safe language with only exception being temporal safety w/out a GC. Rust delivers on that. http://www.adacore.com/uploads_gems/Ada_Safe_and_Secure_Booklet.pdf http://www.adacore.com/uploads_gems/Ada_Safe_and_Secure_Book... Also, SPARK Ada can straight-up prove common errors don't exist in that code with an automated prover. Most engineers would spend quite a bit of time doing that with pencil-and-paper examination or human-driven provers: https://en.wikipedia.org/wiki/SPARK_(programming_language) https://en.wikipedia.org/wiki/SPARK_(programming_language)

nickpsecurity 10y ago

It would be. Especially with mature tooling available from AdaCore. Seem my reply to vertex-four for links containing proof. Here's some examples in industry & CompSci just for kicks: http://www.adacore.com/customers http://www.adacore.com/customers Note that the authors are just doing research on seeing how well Rust handles the problem domain. There's nothing wrong with them avoiding Ada to explore another tool or problem area. They should consider each carefully if they were trying to pick best tool for real work.

naasking 10y ago

> Given the performance and security concerns wouldn't Ada be the natural choice here? Ada would be a good choice, but it's way more awkward to program in. It definitely has some nice features of its own that Rust should steal asap though.