Re: [elpa] externals/emms e9fa4c095d: * emms.el: reverse the order players are collected

[Yoni Rabkin <yoni@rabkins.net>]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

>> --- a/emms.el
>> +++ b/emms.el
>> @@ -1511,7 +1511,7 @@ If the track can be played by more than one player, call
>>      (mapc
>>       #'(lambda (player)
>>  	 (when (funcall (emms-player-get player 'playablep) track)
>> -	   (push player players)))
>> +	   (setq players (append players `(,player)))))
>>       emms-player-list)
>>      (if (< 1 (length players))
>>  	(emms-players-preference track players)
>
> This changes the code's complexity from O(N) to O(N²) in the number of
> players.

Is this because append has to traverse the list all of the way to the
last cdr every time?

> Maybe it's not a very big deal,

Indeed. Since the list of players is a constant (about 7), this ends up
being constant-time. I tend to ignore how short lists are processed, and
I indeed didn't think about big-O of anything when I made this change.

> but the standard way to avoid this is to use `push` inside the loop
> followed by `nreverse` at the of the loop.

Doing it the Right Way (TM) feels better regardless of if it makes any
actual sense(*). Programming is about elegance as much as anything
else. I'll change it to use push and nreverse.


* We've probably spent more time in this conversation than the sum total
  processing time of all of the machines which will ever run this code.

-- 
   "Cut your own wood and it will warm you twice"