From: Simon Strandgaard <0bz63fz3m1qt3001@...>
Date: 2003-06-27T05:18:31+09:00
Subject: Re: collect info about ruby-api

On Thu, 26 Jun 2003 12:17:14 +0200, Tobias Peters wrote:

> On Wed, 25 Jun 2003, Simon Strandgaard wrote:
> 
>> On Wed, 25 Jun 2003 22:57:53 +0900, nobu.nokad wrote:
>> 
>> >   #3   Init_stack(address)
>> >        tell GC the machine stack limit.
>> 
>> I never heard about this before, what does it do ?
>> Can you give me a more lengthy description of it :-)
>> 
> 
> Simon, remember how the GC works: Every time it runs, it marks all objects 
> that are still reachable, from either global variables, or from local 
> variables that are currently somewhere on the execution stack.

Yes I know mark&sweep GC.

 
> Your two methods would declare your variables global.

Yes.   I know there is other methods.
The reason why I suggested these 2 methods is because I want to avoid
confusing people..  No one would have understod me if I had 
presented all posible methods.

These 2 methods represent the only two distinct ways you can tell GC that 
there is a new instance in town.
#1  export a C variable into Ruby.
#2  non-export.. not visible within Ruby.

I have no insigth in what Init_stack does.. I am guessing it belongs to
the second category?

 
> Nobu's suggested to rely on the fact that objects currently referenced 
> from local variables also get marked.
> 
> Local variables are references that live on the stack.
> 
> The Init_stack method corrects ruby's idea of where the C stack starts.
> All objects referenced from the C stack will be marked during a GC run.

I don't know anything about Ruby + stack... tell me more ?

It is not because I don't know what a stack is.. I have done assembler
programming, written compiler, cpu emulator, operating system...  I think
I know what a stack is :-)

I just don't understand the purpose+goal with Init_stack.


>> 
>> It doesn't seems to something public available to the embedding/extension
>> Rubyist's. To me it looks like an internal routine.
> 
> It will be called from ruby_init. You can avoid calling it explicitly if 
> you do all your work from within another function after you called 
> ruby_init, like this:
[snip code]
> Then you can _RELY_ on the fact that _none_ of your objects pointed to by 
> x1, x2, etc, will be destroyed _before_ the do_the_work function returns.

I don't see where Init_stack enters the picture ?

 
> If you can't believe this, then learn about it yourself:

I already know what a stack is.. 
 
 
> And you need an idea of how the garbage collector works. The C code in the 
> file "gc.c" is very readable, if you ignore all the switch/case branches 
> that mention some NODE objects.

I have not read 'gc.c' closely..  I know mark&sweep.
I don't know much about Ruby's internals (NODE ???).


I am still curious..  
* what can Init_stack be used for?
* when is it supposed to be used? 
* any 'short' example of usage?

Im listening :-)

--
Simon Strandgaard