On Jun 5, 2013, at 11:03 AM, Robin Sommer <robin(a)icir.org> wrote:
On Tue, Jun 04, 2013 at 15:24 -0700, you wrote:
(4) bloomfilter_merge(bf1: opaque of bloomfilter,
bf2: opaque of
bloomfilter) : bool
Is this merging bf2 into bf1? Would it make sense to return a new
"opaque of bloomfilter" instead?
From a clarity perspective I think that returning a
bloom filter makes the most sense. From a performance perspective I can see the desire to
do a merge to the left approach that Matthias is proposing.
Alternatively, we might need a
bloomfilter_copy() for the case that one wants to merge without
destroying any of the original filters.
Could we just make the existing copy bif work with it? Now that I know how that's
implemented internally though I'm already skeptical of the idea. :)
International Computer Science Institute
(Bro) because everyone has a network