Re: [xml] Memory reference after xmlFree in XPath evaluation

[Brian Stafford <brian stafford uklinux net>]

(oops, forgot to cc list)
On Fri 11:36,  4 April 2003, Daniel Veillard wrote:
> On Fri, Apr 04, 2003 at 10:59:32AM +0100, Brian Stafford wrote:
> > The problem arises when func() generates a node-set from scratch rather
> than
> > merely referencing nodes in the document being processed.  In this case the
>
> > function will use code like
>
>   First, what version are you using ? CVS ?

Current CVS.

>   Second what function() generate that node set, where do the node
> comes from ?

The function is part of a fairly complex program, unfortunately it's not easy 
to split out on its own to provide as a test case.

> The function is responsible to provide the persistency
> of that set of node.
>
> > as is the case in some libexslt code.  During evaluation of the XPath
> filter
>
>   Which function creates nodes ?

I realised that so I used the implementation of the libexslt tokenize() 
function in libexslt/strings.c from the current libxslt CVS as an example of 
how to manage the persistency.  My function creates a fake node and creates 
the nodeset in exactly the same way.  I will try to write a simple test 
program that reproduces the problem - probably Monday before I have that ready.

> The problem is the lifetime of the nodes.

Agreed

> It's a VERY HARD problem indeed.

Agreed

> Without precise examples it won't get anywhere. If your code creates
> node to return them into node set, it better provide persistency for those.

Are you saying that the structure returned by xmlXPathNewNodeSet() is the 
wrong place to manage persistency of nodes created by XPath function 
implementations?


I will follow up with a test case ASAP

Brian Stafford