On 2010-12-18 <12:35:38>, Hans Hagen wrote:
> On 18-12-2010 12:03, Philipp Gesang wrote:
> >Depending on whether you want to access the non-hashed content as
> >well you might want to use the “next” iterator instead as it’s
> >slightly faster (according to my tests, that is). You won’t get
> >around the local variable, though; I have no clue and no time to
> >check if it’s even technically feasible to iterate a hash table
> >without accessing the hashes.
> 
> indeed, for k, v in next, sometable do ... is faster as it saves one
> function call (i.e. pairs returning the next, table)
> 
> for indexed iteration using for i=1,n do ... is much faster than
> ipairs as it involves no function calls

Well, according to a quick test using “next” with hash tables has
only a *very* slight advantage (in time) over pairs. With arrays
the difference to ipairs() is much higher; in mixed arrays/hashes
it might turn out even higher.

For tables I came up with this relation (“<” meaning less
execution time than):
  while < for < repeat < next < ipairs
where “repeat … until” and “while … do … end” check if the final
index has been reached. But the difference between the three
loops is rather neglegible.

What always baffled me is that in the manual Roberto advertises
ipairs() as the iterator of choice (at least in the v.5.0 doc).
Compared to the other options using it is just, well, erratic.

Regard, Philipp


-- 
()  ascii ribbon campaign - against html e-mail
/\  www.asciiribbon.org   - against proprietary attachments