Hi Stephen,

In using the code with 8.6 you may be entering slightly uncharted waters as
Xerces hasn't to my knowledge been well tested in that weathered
environment. I'm a little curious (I don't remember) about when the HFS+
APIs, including FSpMakeFSRef, were made available, and whether they were
available first in 8.6 or in 9.0.

The code in Xerces 1.6 does a null tvector test in order to try to check for
HFS+ APIs. I'm wondering if this is failing for some reason, and that the
reason you're crashing is that FSpMakeFSRef is simply not implemented in
8.6.

Can you try the following: hack MacOSPlatformUtils.cpp, in the routine
XMLPlatformUtils::platformInit, to simply set gHasHFSPlusAPIs to false and
see if this helps. This should keep the code from calling any HFS+ APIs
including the FSpMakeFSRef you mention.

I'll warn you too, that use of Xerces under systems < 9.0 probably hasn't
been well exercised, and you may run into other problems.

The platform init code in Xerces > v1.6 has been revised to use more Gestalt
based checks for FS functionality, rather than the tvector tests used in
Xerces 1.6 and prior. You might consider getting a recent nightly or cvs
build to see if this fixes your problem also.

I'd love to hear what you find...

James.

On 3/9/02 7:54 PM, "Stephen Marshall" <[EMAIL PROTECTED]> wrote:

> Hi,
> 
> I am having some problems with getting the parser to work with disk
> files on MacOS8.6. The following code works fine up to the point of
> creating the FSRef  for the parent directory:
> 
> DOMParser* parser = new DOMParser;
> const char * fName = "todo.xml\0";
> parser->parse(fName, false); // This is where the code fails
> 
> At this point an illegal instruction exception occurs deep in
> MacOSPlatformUtils.cpp  in the call XMLParsePathToFSRef_Classic(const
> XMLCh* const pathName, FSRef& ref)
> 
> err = FSpMakeFSRef(&spec, &ref); // This is where the code fails
> 
> The spec has been created correctly and points at the correct directory.
> 
> Any suggestions appreciated, including examples of working code that
> I can use to test whether my version of the library has been built
> correctly.
> 
> Cheers
> Stephen


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to