Re: 'in' keyword today (was Re: mathlib)

[Thomas Jericke <tjericke@...>]

On 04/09/2014 10:22 AM, Dirk Laurie wrote:
> 2014-04-09 6:30 GMT+02:00 Coroutines <coroutines@gmail.com>:
>
> > I think I would suggest both an 'in' and 'from' so this would be possible:
> >
> > local a, b, c in some_table
> > local t_concat, t_insert = concat, insert from table
>
> The proposed 'from', on the other hand, is not too hard to imitate
> in pure Lua.
>
>      local t_concat, t_insert = from(table,"concat,insert")
>
> I don't think the impact is dramatic enough to justify another keyword.
I totally agree, a new keyword is not needed, but why not use the same 
keyword for both cases

local a, b, c in some_table
local t_concat, t_insert = concat, insert in table

I am sure the parser will be able to distinguish a RHS "in" from a LHS 
"in".
Actually:
local a, b, c in table
--expands to
local a, b, c = a, b, c in table
--expands to
local a, b, c = table.a, table.b, table.c
-- If "table" is an expression the expression gets evaluated only once.

Could all be done with a preprocessor :-) The last comment only at the 
cost of a local.

local a, b, c = a, b, c in expression
--expands to
local temp_table = expression -- extra local needed
local a, b, c = temp_table.a, temp_table.b, temp_table.c

The next step would be allow functions:
local a, b, c = a, b, c in func, invariant
--expands to
local a, b, c = func(invariant, "a"), func(invariant, "b"), 
func(invariant, "c")

Invariant could be for example a table and the function an additional check:
function get_assert(table, key)
    return assert(table[key])
end
local sinh, cosh in get_assert, require "mathx"

So much fun!

I actually started to analyse the Lua code how I could do a patch for 
this, but I don't have enough time to work on it currently.

--
Thomas