Xavier Leroy
Xavier Leroy
I didn't know about PolymurHash, so thanks a lot for pointing it out to me. The main blocking point in #9764 is not so much the speed of SipHash1-3, but...
Thanks for the reference to the Kundu and Bertino paper, the abstract makes me want to read it. > In this scenario, the current hash function stays for compatibility with...
This is an interesting proposal, so let's keep it alive.
What news? We've known for more than 10 years that Murmur3 suffers from universal collisions (from Aumasson and Bernstein's analysis), see #24 and #9764. Also, the way OCaml uses Murmur3...
Thank you for the concrete example. There's a typo: `Field(y, 0) = y;` should be ` = z;`. More surprisingly, your analysis seems to ignore the "acquire barrier; store release"...
Thanks for the additional explanations. I was too focused on the writing side but now I see the reading side is the concern. If it can help, we could introduce...
> there are situations where it still makes sense to use `Field` Just to make sure we're on the same page: there are two very common such situation, namely 1-...
Excellent! Then I'm in favor of introducing `Load_mutable_field`, documenting it, and using it within the core system (at first).
Nice idea, thanks. Proposed implementation at #554 .
Not sure why I was summoned here :-) I support the use of C99 flexible arrays, it's cleaner and less susceptible to over-optimizations than the size-one-array idiom. Concerning the ocamlyacc...