On Mon, 05 Jun 2006 00:46:12 +0100, Doug Bryant <[EMAIL PROTECTED]>  
wrote:

> ... and thanks for trying!
>
> What are the plans for upcoming memory management in libxml-ruby?  I
> have seen some stuff floating around on the net about the current
> implementation trying to bypass ruby's gc.  Are there plans to use
> rubys mark & sweep api?  Is this related to the bug in question?
>

It may be somewhat related, although there's an aspect I'm missing with  
libxml2, since it's the xmlFreeDoc that actually causes the fault when a  
node exists in two documents. I'm looking more into how xmlDocCopyNode and  
so on works, and I might need to ask on the libxml2 list yet, but I'm  
working on it :)

The current implementation does try to avoid the GC where possible, and  
there was some discussion a while back with Sean about this. He made the  
decision based (I think) on Ruby's performance at the time but with  
hindsight he thought it might be better to change the way memory is  
handled, and stick with the GC where possible. I'm planning a gradual  
migration as stuff needs to be worked on, and I'm hoping to get started  
with the basic XML::Node and XML::Document stuff pretty soon (possibly in  
a cvs branch).

-- 
Ross Bamford - [EMAIL PROTECTED]
_______________________________________________
libxml-devel mailing list
[email protected]
http://rubyforge.org/mailman/listinfo/libxml-devel

Reply via email to