Re: Speed of #t

["Juris Kalnins" <juris@...>]

On Mon, 13 Dec 2010 15:34:02 +0200, Enrico Colombini <erix@erix.it> wrote:

> I wholeheartedly agree with Luiz here: "the array part is just an  
> implementation detail". We should approach from the user's side first  
> and look at internal implications later (if ever).

From the user's side bigger inconsistency is not the #t, which
is very well defined, and _behaves exactly as defined_.
in my opinion, it's this:

print(table.concat{"a","b","c","d", nil, "e"})
./lua: test.lua:1: invalid value (nil) at index 5 in table for 'concat'

Now, manual explicitly states that table value cannot be 'nil'.
Similar gotchas with tables created from return values,
while running ipairs on such tables etc..