←back to thread

Tree Borrows

(plf.inf.ethz.ch)
565 points zdw | 1 comments | | HN request time: 0.215s | source
Show context
jcalvinowens ◴[] No.44513250[source]
> On the one hand, compilers would like to exploit the strong guarantees of the type system—particularly those pertaining to aliasing of pointers—in order to unlock powerful intraprocedural optimizations.

How true is this really?

Torvalds has argued for a long time that strict aliasing rules in C are more trouble than they're worth, I find his arguments compelling. Here's one of many examples: https://lore.kernel.org/all/CAHk-=wgq1DvgNVoodk7JKc6BuU1m9Un... (the entire thread worth reading if you find this sort of thing interesting)

Is Rust somehow fundamentally different? Based on limited experience, it seems not (at least, when unsafe is involved...).

replies(11): >>44513333 #>>44513357 #>>44513452 #>>44513468 #>>44513936 #>>44514234 #>>44514867 #>>44514904 #>>44516742 #>>44516860 #>>44517860 #
Asooka ◴[] No.44513357[source]
While I can't name the product I work on, we also use -fno-strict-aliasing. The problem with these optimisations is that they can only be done safely if you can prove aliasing never happens, which is equivalent to solving the halting problem in C++. In Rust I suspect the stronger type system can actually prove that aliasing doesn't happen in select cases. In any case, I can always manually do the optimisations enabled by strict aliasing in hot code, but I can never undo a customer losing data due to miscompilation.
replies(2): >>44513415 #>>44515614 #
1. immibis ◴[] No.44515614[source]
> which is equivalent to solving the halting problem

Most worthwhile undefined behaviour is. If the compiler could tell whether it was happening or not, it would be defined as an error. Surely detecting whether a tree borrow violation happens is also equivalent to solving the halting problem?