4 ms·

This particular linux thing wouldn't involve rewriting anything in Rust.

cryptonector 7y ago
I don't know why you got downvoted. This is about new drivers. I suppose some existing drivers might later get rewritten, but that's later, when there's been more experience with it. In any case, FF shows that rewriting a complex C/C++ codebase in Rust bit by bit is feasible, though again, that doesn't seem to be the intent in this case.
- anaphor 7y ago
  Correct me if I'm wrong, but these drivers wouldn't be merged into the kernel tree itself, right? They would be written as modules. I doubt any of the core developers would feel comfortable reviewing Rust code at this time.
  - nwmcsween 7y ago
    About zero chance of rust drivers being merged into the kernel now or in the near future.
- notyourday 7y ago
  FF shows that new language advocates are as usual out to lunch. It is as if everyone forgot the time when the Netscape ruled the web but some clowns decided they want to rewrite it and make it more awesome driving its market share to nearly zero and company out of business. Mozilla should have continued to re-iterate on FF code, slowly but surely improving it rather than going "Oh look, we can make it soooo much better with this new awesome language" P.S. The really funny part is that it is basically the exactly same project that is engaging in another "rewrite" just with different corporate owners.
  - jcranmer 7y ago
    The Netscape rewrite was a complete tossing of the codebase, starting from scratch on virtually every component of the web browser and internal implementation details. It wasn't the start of the troubles, but it did make things worse. Mozilla isn't doing that today. The current code is being "slowly but surely improved," and the use of Rust is largely happening only in new components (and not all new components are being written in Rust!). There's very little gratuitous conversion of C++ code to Rust.
    - notyourday 7y ago
      > Mozilla isn't doing that today. The current code is being "slowly but surely improved," and the use of Rust is largely happening only in new components (and not all new components are being written in Rust!). There's very little gratuitous conversion of C++ code to Rust. So one can build FF without Rust? Because if the answer is "Not any more" then it is a beginning of a rewrite, just done slower.
      - jcranmer 7y ago
        It is not the beginning of a rewrite just because Firefox requires Rust. All that it means is that there are now three languages you can choose to use to implement Firefox components, instead of just two. You're conflating the ability to choose Rust in addition to C/C++ with the intent to replace C/C++ code with Rust.
        notyourday 7y ago
        Not at all. There was a state X, where to build FF one needed to use a certain tool chain. The new state is Y. To build FF one now has to use two tool chains: the previous one and the new one. Not having either of those tool chains prevents the build. That's a definition of a beginning of a rewrite. And it is an incredibly dumb idea from a business perspective which will bite Mozilla in the ass in the future because it is increasing the build complexity today and tomorrow for some pie in the sky promise that may or may not pan out in future. I'm not saying "Don't use Rust for a new project". I'm saying that the software industry is littered with the corpses of entities that decided to add a new way of doing something to an existing project where it did not pan out.
  - Bjartr 7y ago
    IIRC, Rust came about because Mozilla did took the iterative improvement approach with it's DOM infrastructure code and after multiple tries that all involved critical, security impacting bugs being missed, no matter the level of diligence they required from contributors, they wanted to see if they could create a system that enforced that diligence up front. They chose to create a language whose semantics and defaults make it a lot harder to make the kinds of mistakes that kept dining previous attempts at iterative improvement. To that end, Rust has been, in practice, for the Mozilla team, the most effective way to achieve their goals. Whatever else there is to say about Rust, it did solve a real problem Mozilla had, that they had been otherwise unsuccessful at solving after multiple serious attempts over a period of years.
    - notyourday 7y ago
      It did not solve a real problem. It may be a path towards a solution of a real problem that Mozilla has but it certainly has not solved it. What it did do is increase a complexity by adding another toolchain into the build process. Not only that, it added a new, not stable tool chain that impacted how others can build FF. And it has not been clear what so ever that this has delivered a major improvement at the cost of added complexity.

I don't know why you got downvoted. This is about new drivers. I suppose some existing drivers might later get rewritten, but that's later, when there's been more experience with it. In any case, FF shows that rewriting a complex C/C++ codebase in Rust bit by bit is feasible, though again, that doesn't seem to be the intent in this case.

anaphor 7y ago
Correct me if I'm wrong, but these drivers wouldn't be merged into the kernel tree itself, right? They would be written as modules. I doubt any of the core developers would feel comfortable reviewing Rust code at this time.
- nwmcsween 7y ago
  About zero chance of rust drivers being merged into the kernel now or in the near future.
notyourday 7y ago
FF shows that new language advocates are as usual out to lunch. It is as if everyone forgot the time when the Netscape ruled the web but some clowns decided they want to rewrite it and make it more awesome driving its market share to nearly zero and company out of business. Mozilla should have continued to re-iterate on FF code, slowly but surely improving it rather than going "Oh look, we can make it soooo much better with this new awesome language" P.S. The really funny part is that it is basically the exactly same project that is engaging in another "rewrite" just with different corporate owners.
- jcranmer 7y ago
  The Netscape rewrite was a complete tossing of the codebase, starting from scratch on virtually every component of the web browser and internal implementation details. It wasn't the start of the troubles, but it did make things worse. Mozilla isn't doing that today. The current code is being "slowly but surely improved," and the use of Rust is largely happening only in new components (and not all new components are being written in Rust!). There's very little gratuitous conversion of C++ code to Rust.
  - notyourday 7y ago
    > Mozilla isn't doing that today. The current code is being "slowly but surely improved," and the use of Rust is largely happening only in new components (and not all new components are being written in Rust!). There's very little gratuitous conversion of C++ code to Rust. So one can build FF without Rust? Because if the answer is "Not any more" then it is a beginning of a rewrite, just done slower.
    - jcranmer 7y ago
      It is not the beginning of a rewrite just because Firefox requires Rust. All that it means is that there are now three languages you can choose to use to implement Firefox components, instead of just two. You're conflating the ability to choose Rust in addition to C/C++ with the intent to replace C/C++ code with Rust.
      - notyourday 7y ago
        Not at all. There was a state X, where to build FF one needed to use a certain tool chain. The new state is Y. To build FF one now has to use two tool chains: the previous one and the new one. Not having either of those tool chains prevents the build. That's a definition of a beginning of a rewrite. And it is an incredibly dumb idea from a business perspective which will bite Mozilla in the ass in the future because it is increasing the build complexity today and tomorrow for some pie in the sky promise that may or may not pan out in future. I'm not saying "Don't use Rust for a new project". I'm saying that the software industry is littered with the corpses of entities that decided to add a new way of doing something to an existing project where it did not pan out.
- Bjartr 7y ago
  IIRC, Rust came about because Mozilla did took the iterative improvement approach with it's DOM infrastructure code and after multiple tries that all involved critical, security impacting bugs being missed, no matter the level of diligence they required from contributors, they wanted to see if they could create a system that enforced that diligence up front. They chose to create a language whose semantics and defaults make it a lot harder to make the kinds of mistakes that kept dining previous attempts at iterative improvement. To that end, Rust has been, in practice, for the Mozilla team, the most effective way to achieve their goals. Whatever else there is to say about Rust, it did solve a real problem Mozilla had, that they had been otherwise unsuccessful at solving after multiple serious attempts over a period of years.
  - notyourday 7y ago
    It did not solve a real problem. It may be a path towards a solution of a real problem that Mozilla has but it certainly has not solved it. What it did do is increase a complexity by adding another toolchain into the build process. Not only that, it added a new, not stable tool chain that impacted how others can build FF. And it has not been clear what so ever that this has delivered a major improvement at the cost of added complexity.