From: "daz (c)" <dooby@...10.karoo.co.uk>
Date: 2007-12-14T20:29:50+09:00
Subject: Re: IRHG - PUSH_FRAME?

Charles Thornton wrote:
> The following define statement is driving me crazy
> I have never seen a  #define used this way!
> 
> Can someone explain what is going on here?
> 
> Does this somehow create a new frame without
> allocation, and add it to a ruby_frame linked list?
> 
> #define PUSH_FRAME() do {        \
>    struct FRAME _frame;        \
>    _frame.prev = ruby_frame;        \
>    _frame.tmp  = 0;            \
>    _frame.node = ruby_current_node;    \
>    _frame.iter = ruby_iter->iter;    \
>    _frame.argc = 0;            \
>    _frame.flags = 0;            \
>    _frame.uniq = frame_unique++;    \
>    ruby_frame = &_frame
> 
> #define POP_FRAME()              \
>    ruby_current_node = _frame.node;    \
>    ruby_frame = _frame.prev;        \
> } while (0)
> 
> 

C90 compilers won't allow a declaration (struct) amid code,
so it's wrapped in a run-once-only block "do {...} while(0)".
The declaration is at the top of the new block so a compiler error
is avoided.
For some reason just "{...}" doesn't work in all cases, so it
was changed in rev 2401:
http://svn.ruby-lang.org/cgi-bin/viewvc.cgi/branches/ruby_1_8/eval.c?view=diff&r1=2400&r2=2401

Subsequent use of POP_FRAME() closes the block.
They are two distinct macro definitions although POP_FRAME()
might look embedded to the human eye. ;)
(Absence of backslash marks the end of a macro.)

daz