string.sub and string.byte: unspecified behaviour or manual underspecification?

[Lorenzo Donati <lorenzodonatibz@...>]

Hi all!

I found two corner cases for string.sub and string.byte and I'd like to 
know if their behaviour is something we can rely on or it is unspecified 
(the manual says nothing about it).


(1) -- string.sub(s,i1,i2) --

When both indexes are greater than string length the result is the empty 
string. When i2 is greater than the length, it is treated as the length.

e = ""
s = "abc"
assert( s:sub(5,5) == e )
assert( s:sub(5,10) == e )
assert( s:sub(1,10) == s )


(2) -- string.byte(s,i1,i2) --

When the string is empty, the returned value is `nil`. When the indexes 
are "out of range" (as the case (1)), the returned value is `nil`.

assert( e:byte() == nil )
assert( s:byte(5,5) == nil )
assert( s:byte(5,10) == nil )
assert( table.concat( {s:byte(1,10)}, " " ) == "65 66 67" )



The aforementioned behaviour is very intuitive and convenient, but the 
manual (5.1.4 and 5.2 beta alike) doesn't specify it. Is it unspecified 
behaviour? If not, I think the manual should report it.


Cheers.
-- Lorenzo