On Jul 8, 2018, at 14:11, David Kastrup <[email protected]> wrote:
> 
> I think it's for providing a definite point of time for running the
> finishing hooks of a context's iterators and making the context
> unavailable for \context ... [= ...] continuations.
> 
> Garbage collection is not determinate.


I think I see.  If a reference-holder is expected to drop its reference at a 
specific point, then if the reference is still held at the time of destruction, 
it is a programming error.  Whether to reduce the reference count at 
destruction or leave it permanently above zero is then just a question of 
robustness given the failure of the reference-holder to follow protocol.
— 
Dan

_______________________________________________
lilypond-devel mailing list
[email protected]
https://lists.gnu.org/mailman/listinfo/lilypond-devel

Reply via email to