Andreas Rumpf
Andreas Rumpf
In reality moving code around makes the problem worse -- we tried that with nim-lang/zip etc and more recently with 'fusion'. You need to ensure that the CI covers the...
> Do we make a json3that is 90% identical to json2 ? If only there was a way to re-use code via templates, Nim's import-export mechanism, `.deprecating` individual procs rather...
They are tested yes, but only a single specific commit/version of every important package (well, usually "git master").
> Sad to read that. Why not have some kind of "legacy" mode for code that can't be updated? Sure but we need to be able to write code that...
> And is it the goal to have Nil-safety as the default in the long run? Yes, sure.
I think we continue to need `unchecked ref/ptr T` for good but we can write it with the explicit `unchecked` in the long run.
It's already available via `--experimental:typeBoundOps`.
> How much harder is it to destroy at last usage? Not much harder but I have never implemented it so I could be wrong. Also, the resulting edge cases...
> With the new architecture described above, is there any realistic chance that there will be a new backend that targets the Rust compiler? Compiling Nim or NIFC to Rust...
Well Nim does elide wasMoved+destroy pairs so it's quite comparable and should also enable more tail recursions. But tail recursion should have a dedicated syntax as relying on it as...