Re: Lua 5.1 (work1) now available

[Philippe Lhoste <PhiLho@...>]

Eric Tetz wrote:
> Roberto Ierusalimschy wrote:
>
> > - there are several other small changes
>
> What I've found (by comparing _G in 5.0 vs 5.1):
>
> Removed:
>    io.popen
>    arg.n
>
> Added:
>   _PATH
>      New name for LUA_PATH (now deprecated) used by 'require'
>
>   string.reverse (s)
>     returns a reversed copy of string 's'
[snip]

This one is very strange...
Can somebody give me a real-world use case where this function is so 
useful that it needs to go in a standard library? I can't think any 
practical use myself. Perhaps it is a temporary test function?

I would rather see a string buffer library added to the standard 
library. One that would allow building large strings without hashing and 
perhaps GC overhead, allowing direct read/write of bytes in arbitrary 
addresses within the string size, automatic growth of buffer (or not), etc.

Unless somebody really convinces me that the table approach is almost as 
memory efficient and hashing can't be avoided anyway because Lua does it 
on every string it sees...

(If I do:
x = string.sub(largeStr, i, j)
f:write(x)
x = nil
does x is being hashed?)

I dream of such library for quite a long time (since Roberto did his LTN 
about string concatenation efficiency) and discovered later that it 
exists in Java (and probably on a number of other languages with high 
level string processing).

I would write it myself, but never got time to start it...
So would it be useful at all?

--
Philippe Lhoste
--  (near) Paris -- France
--  Professional programmer and amateur artist
--  http://Phi.Lho.free.fr
--  --  --  --  --  --  --  --  --  --  --  --