Wes Brown

@tonsky Interesting point, I hadn't realized that. What is the impact of layout on a tuple join? I hadn't realized that there was an implicit requirement, which I will try...

@tonsky So, I was thinking about it. We could put a conditional that measures the number of attributes given. If `l1` or `l2` exceeds 8, then we set `acopy?` to...

@tonsky Fair points about not relying on implementation details of the platform. I've been examining and thinking about the logic here: * We do an `acopy` if and only if...

@tonsky I expanded the scope of my thinking and examined the `Relation` records. Your comment about using 'proper data structures' nudged me along that path. Couple suggestions: * `array-map` ordering...

@tonsky Would you be OK with introducing `frankiesardo/linked`? It's a fairly nice ordered map implementation for both CLJ and CLJS. https://github.com/frankiesardo/linked

@tonsky We could. Rather than use a hash map to contain the relation to key index, we could use a vector instead in `:attrs`. Unless we're joining very large numbers...

I can attest to the query engine working well when the `IDB` and `ISearch` protocols are implemented. However, in that case, it does one pattern at a time, which can...

@thedavidmeister An alternative is to hash the squuid into a `2^53` bit number, but depending on how many entities you need to be tracking, you risk collision with a 53...

I think this may be biting us for an MacOS X build: ``` ../../go/src/github.com/google/gopacket/pcapgo/capture.go:46:46: ETH_P_ALL not declared by package syscall ```