[cfe-dev] RFC: change string hash function in libc++

me22 me22.ca at gmail.com
Mon Dec 5 22:33:39 PST 2011


On Sun, Dec 4, 2011 at 09:50, Howard Hinnant <hhinnant at apple.com> wrote:
> One of the decisions I'm making based on this test is what to do with two (or four) size_t's when I need to combine them into one.  What's checked in is:
>
>        return __u.__a ^ __u.__b;
>

I'd encourage you to pick something that's not reflexive, and doesn't
always give 0 for (i,i).

I'm picturing something like a sparse matrix where most entries are on
the diagonal.

~ Scott




More information about the cfe-dev mailing list