switch / if-elseif-elseif-else term

[danielmewes]

Currently the r.branch() term isn't that great to use if you have more complex case distinctions.

Proposed by @wojons: Add a variadic term taking 2N+1 arguments

r.switch(cond1, val1, cond2, val2, ..., condN, valN, default_val)

Its semantics are those of the following code block in C:

if (cond1) return val1;
else if (cond2) return val2;
...
else if (condN) return valN;
else return default_val;

We could also make r.branch() variadic, instead of introducing a new term.

[wojons]

@danielmewes

Thank you, One thing i want to add. I think it would be important for that after the default value you can submit another aarray or object that allows u to control if its a if else or if its a switch. Or a r.break term will need to be added so it will stop letting things fall though.

[timmaxw]

I think that we should simply allow r.branch() to take 2N+1 arguments where N>=1. The current behavior of r.branch() is simply the N=1 case.

[wojons]

i think thise is ur guys call on how you want to do this.....

[gchpaco]

As far as bike shedding goes, multi-branch if statements mean "cond" to my brain. I could live with switch. Branch is a bit weird, but I could probably get used to it in time.

[timmaxw]

I don't really like the name r.branch either (I wanted to call it r.if_then_else) but given that we already have the command r.branch, I think we shouldn't introduce a new separate command.

[mlucy]

Another vote for making branch more powerful rather than adding another term. We already have so many that it's hard to keep track of them all.

[danielmewes]

Let's finalize this during the 1.17 ReQL discussion period, not now...

[wojons]

The only reason that i would recommend r.switch over r.branch is as a user first looking at rethinkdb and trying to find a way to do condations r.branch does not sound like its going to be for what i amlooking for i think it was @AtnNn that showed me that command. But as a devloper seeing a command called switch sounds like a switch statment.i would ad r.switch and deperatcate r.branch.

[coffeemug]

:+1: for extending r.branch and not introducing a new term.

[deontologician]

extending r.branch seems obviously good.

[wojons]

I think r.branch sounds more like its forking something but fair enough for extending it as long as i dont have to use 10 of them.

[VeXocide]

Extending r.branch seems obvious, are we considering renaming it as well?

[danielmewes]

I'd like to keep the renaming discussion separate and for later.

[danielmewes]

Marking as settled for making branch variadic to support multiple branches.

[pilwon]

:+1: fantastic idea to make r.branch more powerful!

[encryptio]

Done in next as of bf5b28b, CR 3201

[mlucy]

@encryptio -- is there a docs issue?

[encryptio]

@mlucy Writing an issue out now.