My goal is that this should be handled by cbdata checks, but I have not
yet found a foolprof way of doing that (at least not until all data has
reference counters)
/Henrik
adri wrote:
> * add some REAL tasty logic in ncomm_close(), which is called by
> comm_close() when a fd goes away. Right now, I'm walking the callback
> list and completing any pending callbacks. If one of the callbacks is
> a pending write, we close it with COMM_ERR_CLOSING so the routine
> knows we're going away. The logic is *TASTY* here, and its not going
> to scale well because we are searching the pending request list, but
> again this code will only exist whilst small debugging is done, and then
> it'll come out and be replaced with something a little nicer.
Received on Fri Feb 16 2001 - 14:08:03 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:13:32 MST