Re: question about internals

Alle 20:32, giovedì 18 marzo 2004, hai scritto:
> > The problem is quite tricky.
>
> indeed!!! I thought the times of max-6-chars-for-identifier languages
> were over, but I was wrong :-(

unfortunately not. In scientific research this is still persecuted. i'd like 
to say a lot about this, but this is not the place. However, there's interest  
about a fortran DOM reader because lots of fortran programmers need to access 
xml (as i'm used to say, they wake up a morning, and they discovered that the 
outer world had not slept at all) and the currently available parsers are SAX 
and written entirely in F90. 

> you're right. At the risk of being obvious: as long as a Gdome2
> structure for a node x stays in memory, no matter _how_ you access node
> x, you'll get the same pointer to the same Gdome2 structure for x. If
> you unref the structure and the ref counter gets to zero, the structure
> is freed and the next time you ask for it it'll most likely have a
> different address.

good, this gives me interesting conclusions about the new version of my 
library. I think this cache structure I implemented lacks some formal 
requirement, although is needed because there's no way to free the nodes (and 
whichever structure you use for non atomic behaviours) once the file is 
closed. Moreover, i can rely only on integers (due to the lack of 
standardization in fortran about the notion of pointer), so i need it also as 
a translation between an integer and a C pointer and viceversa.

> > last question: if i obtain a nodelist, when unref'ing it the referred
> > nodes are automatically unrefferred (refCount--) ?
>
> yes, you don't have to worry about (neither you know) what the NodeList
> refers.

good. Many thanks.