Re: Some crazy proposals :-)

[Marc Balmer <marc@...>]

Am 09.07.15 um 19:13 schrieb Andrew Starks:
> 
> 
> On Thursday, July 9, 2015, Rena <hyperhacker@gmail.com
> <mailto:hyperhacker@gmail.com>> wrote:
> 
>     On Jul 9, 2015 11:04 AM, "Egor Skriptunoff"
>     <egor.skriptunoff@gmail.com
>     <javascript:_e(%7B%7D,'cvml','egor.skriptunoff@gmail.com');>> wrote:
>     >
>     > Hi!
>     >
>     > The following "proposals" are just for fun.
>     >
>     > -----------------------------
>     > 1) A must-have "dont" statement :-)
>     >
>     > dont
>     >    -- code inside dont..end block will be skipped
>     >    -- useful for easy commenting do..end block
>     > end
>     >
>     > "dont" is also applicable to any block: if, while, for, multiline
>     assignment, function definition, etc.
>     >
>     > dont if ... then -- this condition will not be evaluated
>     >    -- this code will be skipped
>     > else
>     >    -- this code will be skipped too
>     > end
>     >
>     > "dont" statement acts as --[[...]] comment, but only for a single
>     statement inside
>     > When using "dont" we do not need to search for a place where very
>     long statement ends, so using "dont" is more easy than --[[...]]
>     > The "dont...end" is a syntactic sugar for "dont do...end"
>     >
>     > -----------------------------
>     > 2) Famous "comefrom" operator as the opposite to "goto"
>     >
>     > ...
>     > ::label1::
>     > ...
>     > comefrom label1, label2, label3
>     > ...
>     > ::label2::
>     > ...
>     > ::label3::
>     > ...
>     >
>     > is equivalent to
>     >
>     > ...
>     > goto label0
>     > ...
>     > ::label0::
>     > ...
>     > goto label0
>     > ...
>     > goto label0
>     > ...
>     >
>     > -----------------------------
>     > 3) Labels are considered to be first-class citizens
>     >
>     > ::Label1::
>     > ...
>     > local my_label = Label1  -- assigning a value of type "label" to a
>     variable
>     > ...
>     > -- Now "goto" operator gets its real power!
>     > goto my_array_of_labels[where_do_you_want_to_go_today]
>     > ...
>     > -- And of course the same is true for "comefrom" :-)
>     > comefrom get_some_labels()
>     > -- it is equivalent to "comefrom label1, label2,..."
>     > -- where label1, label2,.. are values returned by LAST invocation
>     of get_some_labels()
>     > -- "comefrom" without labels is valid (it just does nothing useful)
>     > ...
>     >
>     > -----------------------------
>     > 4)  Program can modify its own body using methods of labels
>     > A label is a "bookmark" between statements in a program (or before
>     first statement or after last statement)
>     > A label may be named (defined at compile time) or unnamed (created
>     at runtime)
>     >
>     > ...
>     > some_func(some_params)
>     > ...
>     > -- standard library function to create unnamed label just before
>     current statement
>     > local here = labels.this_label()
>     > -- create label just before some_func invocation
>     > local previous_some_func_invocation =
>     here:search_nearest("backward", "call", "some_func")
>     > -- replace "some_func()" with "another_func()" in the program body
>     > previous_some_func_invocation:replace_statement("another_func()")
>     > ...
>     > -- go to unnamed label
>     > goto previous_some_func_invocation
>     > ...
>     >
>     > -----------------------------
>     > The source of craziness:
>     > www.modell.com/Magery/SPharmful.html
>     <http://www.modell.com/Magery/SPharmful.html>
>     >
>     > -- Egor
>     >
> 
>     Computed goto isn't *that* crazy. It's how things like jump tables
>     and switch statements are implemented. Lua doesn't really *need*
>     them though since it has first class functions.
> 
> 
> I will not rest until we can use Lua patterns for variable names:
> 
> .*_foo = nil
> 
> Ruby probably already has this feature.  

Perl for sure has, but it is not readable...