[Xapian-discuss] Ruby Bindings

[James Aylett]

On Sun, Sep 13, 2009 at 01:15:11PM +0100, Olly Betts wrote:

> > In 1.0, Weight is only subclassable in Python.
>
> While there's a conditional for Python there in 1.0, it is actually
> commented out so really no different to 1.1.

Oops, missed the comment. My fault for only looking at it in less.

> > In 1.1, this is disabled entirely. I can't find a ticket for this, so
> > I don't know what the current discussion is around this.
> 
> The problem was that SWIG didn't wrap it properly due to handling of
> the private clone method.  It's quite possible that issue is now fixed
> in SWIG, and it may not be an issue for 1.1 anyway as the Weight API
> has been overhauled.

That's (probably) no longer a problem, but unserialise() returning a
pointer is raising warnings that this is considered
'dangerous'. Ignoring this for the time being, there are then a whole
load of errors for Weight member function wrappers to do with
threading. This all with 1.3.40, so it doesn't look good unless
someone has time to look at this properly.

J

-- 
  James Aylett

  talktorex.co.uk - xapian.org - uncertaintydivision.org