Re: Why are Lua Modules Only a Runtime Component of the Language?

[<jgiors@...>]

> -------- Original Message --------
> Date: Mon, 7 Mar 2011 09:19:27 -0600
> From: Paul Bonser <misterpib@gmail.com>
> Subject: Re: Why are Lua Modules Only a Runtime Component of the
> 	Language?
> To: Lua mailing list <lua-l@lists.lua.org>
> Message-ID:
> 	<AANLkTi=AcZLCBA-wxOjUaQcPj0pkJOQZ69x0FocuRebu@mail.gmail.com>
> Content-Type: text/plain; charset=ISO-8859-1
> 
> On Mon, Mar 7, 2011 at 9:06 AM, Jon Akhtar <akhtar@mindspring.com> wrote:
> > Is there a compelling reason for not making 'require' and 'module'
> > keywords in the language.
> 
> Is there a compelling reason TO make them keywords?
> 
> >
> > Since I am working on a static analysis tool for Lua, handling Lua
> > modules is something I need to do. In doing so, I find myself
> > continually nagged by this question of why modules have a second class
> > status. Why is 'require' merely a function with the name 'require'?
> 
> For the same reason that classes aren't "first-class" in Lua, I'd say.
> 
> >
> > The only reasons I can come up with is that modules (like almost
> > everything in Lua) are optional, and may not be included in embedded
> > Lua, and that they were built using lower level api's like setfenv.
> 
> Right, it seems to be the "Lua way" to include a simpler, lower-level
> functionality that allows you to implement specific, higher-level
> functionality, but not to force any specific implementation on you.
> 
> >
> > That is fair enough. I was just wondering if adding the module system
> > as a language (grammar) feature vs a library had been considered and
> > if anyone had any additional insight on the matter.
> 
> It's part of the standard library, which is implemented in C. I would
> say keeping the grammar as simple as possible is a pretty valuable
> goal.
> 
> Why add something to the grammar when it works fine as a function? For
> static analysis just assume that when someone says "require" or
> "module" that they are using the standard functions. If they aren't,
> they either know what they are doing or are asking for trouble anyway,
> right? :P
> 
> >
> > Thanks In Advance,
> >
> > Jon Akhtar
> >
> 
> -- 
> Paul Bonser
> http://probablyprogramming.com
> http://twitter.com/pib

+1 all of the replies

In addition, it is nice that it is possible to do things like the
following:

--[[]]  -- Note: Untested code snippet for expository purpose only.
--[[]]  -- Re-load module "myModuleName"
--[[]]  package.loaded[ "myModuleName" ] = nil
--[[]]  require "myModuleName"

Also, if require were changed into a keyword, what would that mean for
package.path? Would it be converted to a function? Would that function
also be a keyword? What about cpath?

It starts getting pretty ugly, though I suppose the data members of the
"package" table could continue to be supported as-is, even if "require"
and "module" became keywords. However, it seems strange (at least to me)
to have keywords whose behavior depends on runtime variables stored in a
global table.

John Giors
http://ThreeEyesSoftware.com
jgiors@ThreeEyesSoftware.com