Rename `shift!` and `unshift!` ?

[c42f]

As mentioned on slack a few days ago, unshift! may be inherited, but it's a reasonably poor name IMHO. As pre-1.0 is the time for fixing these things, how about it?

Since prepend is taken, how about unshift!->pushfront! and shift!->popfront!?

[StefanKarpinski]

We could do a wholesale rename of these functions:

old	new
push!	rpush!
pop!	rpop!
unshift!	lpush!
shift!	lpop!

[bramtayl]

Or a left/right (start/end? first/last?) keyword if performance issues could be worked out

[Sacha0]

Left and right seem a bit ambiguous; perhaps @c42f's front/back or something similar would be clearer?

[StefanKarpinski]

I'm not sure that front and back are clearer than left and right. Of course, for left/right terminology it's a bit unfortunate that we print vectors vertically, but front/back aren't better in that respect.

[StefanKarpinski]

Oh, and we already use l and r for left and right in reductions, so there's precedent – unless we want to change those too.

[Sacha0]

I'm not sure that front and back are clearer than left and right.

I'm not sure either. front associates with lower indices (like Base.front) and back with higher indices in my mind, but I am sure that sort of association varies.

Perhaps better ideas: head-tail and first-last seem distinctly less ambiguous given their preexisting meanings?

[StefanKarpinski]

The names hpush!, hpop!, tpush! and tpop! aren't bad. However we already associate left and right with the ordering of arrays in foldl and foldr, etc. Actually, since those have the l and r at the end, the analogous names are pushl!, popl!, pushr! and popr! which I kind of enjoy :)

[c42f]

Out of the options so far for renaming all four functions, the suffixed versions pushl!, popl!, pushr! and popr! look best to me.

I have some reservations about renaming push! and pop! though. For example, consider pushing and popping to a priority queue. In this case there's no "right" and "left", there's only a natural ordering and a data structure which returns the least/greatest element. There's also the version of pop! which takes a key for use with dictionary like data structures, and popr! would also not make much sense as a verb in that case.

[quinnj]

I think push! and pop! can stay and be used by PriorityQueues. They can also just alias pushr! and popr! for Arrays.

[moldray]

I love shift! & unshift! , it is the same with javascript. please reserve this!

[KristofferC]

So far, I like the first post's suggestion the best.

[stevengj]

I think it's good to use push*! and pop*! names to emphasize the similarity between the variants, and so that the variants can be found easily by tab completion. (I like the r and l suffix suggestion.)

[JeffBezanson]

-2 to r and l terminology. The names are a bit ugly, and there are 4 of them. I would rather only rename shift and unshift like the OP suggests. But I also feel shift and unshift are used widely enough that we might as well just keep them. It would be unfortunate to introduce terms nobody else uses for such incredibly common operations.

[c42f]

@JeffBezanson C++ vector uses push_front, so there is some precedent for this.

I assume the existing names can be traced back to the shift builtin in sh, and presumably perl picked this up and invented unshift (http://www.perlmonks.org/?node_id=613144) Other languages (php, javascript, ...?) followed so there's plenty of precedent.

[cormullion]

I prefer proper English words... :) (unshift is in the OED dating back to 1972.)