On Fri, 2007-09-21 at 12:20 -0700, Erick Tryzelaar wrote:
> On 9/21/07, skaller <[EMAIL PROTECTED]> wrote:
> > At present every Felix 'file' has to #import <flx.flxh>.
> > To get rid of this, I'm thinking out loud as usual.
> >
> > Felix has 4 inclusions:
> >
> > #include
> > #import
> > include directive
> > compiler accepts multiple file names
> >
> > #include makes #directives, including #macros, local
> > to the file and isolated from the #including file.
> > It neither imports nor exports any #macros.
>
> Now that we have scoped syntax extensions, I'm not sure if macros are
> as important as they were before. We could just drop the #include
> since I bet they'll be less name collision.
We could, although #include is logically the equivalent of
include at the # level: each #include file is independently
lexed, and so the token stream produced can be cached
to avoid lexing it unless it is changed.
If we go to a unicode (32 bit char) lexer this could be
important.
It's the principle of the thing, that you can split a file
into includes, just as you can split a function into smaller
ones. However
#define ..
line1
line2
has to be split into
#include 1
#Include 2
#define
line1
#define
line2
i.e. the #define has to be duplicated, just as when you factor
a top level function into several top level functions you have
to pass the original function's context to each sub-faction
separately.
What Python does might be interesting model:
Directory X:
File _init.flxh
File A
File B
--> #include X/A imports A with X._init glued in the front.
so you can define shared stuff in _init.flxh. Then if you
say
#include X
you get the lot. But we might want to use the packaging idea for
includes, not #includes .. ouch.
> We should keep a flag to flxg that will turn off automatic importing
> of flx.flxh though, in case someone wants to write a completely new
> syntax.
It's the other way: flxg will require
flxg --import=flx.flxg std xxx
The 'flx' script will default this, so 'flx' will need
the flag to turn it off.
The flag has been implement for eons and aeons .. it just
doesn't do anything (it's actually called 'hash_include'
at the moment).
--
John Skaller <skaller at users dot sf dot net>
Felix, successor to C++: http://felix.sf.net
-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Felix-language mailing list
Felix-language@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/felix-language
