Re: gvariant threading snafu ?

[Ryan Lortie <desrt desrt ca>]

Hi Michael

On Fri, 2010-02-26 at 10:02 +0000, Michael Meeks wrote:
> 	This is a similar situation; interning ref-counted strings, whose
> references are held in noddy hand-coded hash-table (to ensure we have a
> unique version):

I was thinking of doing something similar but I took a pass on it.  I'd
have had to remove the refcount>0 check from _ref() in order to allow
this sort of 'bring it back from the brink of death' behaviour and then
I'd also have to worry about table reinsertion, etc.

I'm gonna stay with the way it is now since it (hopefully) works.


Cheers