Re: [xml] xmlCleanupParser() question / patch

["Martin B." <0xCDCDCDCD gmx at>]

Daniel Veillard wrote:
> On Tue, Jan 19, 2010 at 07:54:20AM -0700, James Hart wrote:
> (....)
> > For example, in our system we have specifically used a RAII design
> > pattern so multiple libraries can access libxml and the initialization
> > and cleanup of the library is handled by a reference counting systems
> > because we have no control of the order these libraries will be loaded
> > in our system.  The goal we have is to completely unload and cleanup
> > the library when there are no libraries requiring it's use anymore and
> > to reload it if a library requires it again.  If the API changes
> > functionality, yet is link compatible it is very possible the change
> > you proposed will go unnoticed for some time.
>
>   suffice one of the shared library used by your system start using
> libxml2 under the hood, say to parse a configuration file, suddenly
> everything breaks. The problem is that this is safe to call only in a
> fully controlled environment, for example in xmllint .

Hmm ... but this point of view would mean that the whole mechanism for 
this function is bad for a shared-lib environment, no matter how it's 
called.

So I could think of this:
* xmlInitParser() and xmlCleanupParser() get deprecated (via the various 
compiler specific attributes)
* xmlInitParser() is left the same as now
* xmlCleanupParser() is made a noop
* New Functions xmlInitLibrary() and xmlCleanupLibrary() that have one 
additional treat: They maintain a counter, so that each call to 
xmlInitLibrary() must be matched by a call to xmlCleanupLibrary() and 
only if the counter reaches zero in the cleanup function is the cleanup 
actually performed. This way the shared lib problem would go away if all 
shared libs are well-behaved.

Would this make sense?

br,
Martin