request structures yes, and what you say about refcnt is again correct.
default conn however was designed never to be freed and RECONNECT flag
is sorta hack to replace it. so my concern in this case is that we have
to make sure that both request and result code flows where we deal with
connection going dead do indeed deal with this correctly in all cases
when we deal with default connection. please file a bug and i will be
able to look closer at it a wee bit later or otherwise Mark or Rich may 
find some time to evaluate it.

BeeOnRope wrote:

> Then the refcount will be only 1 since no requests refer to it, and
> the connection will be cleaned up in the simple_bindifnot_s even
> without force, since refcount drops to zero.
> 
> My assumption is that all request structures must eventually be
> cleaned up, or else there is a big leak there, and when a request is
> cleaned up, the corresponding connection always has its refcount
> decremented.
> 
> In short, ISTM that the refcount mechanism works, and it's not clear
> to me how the simple_bindifnot_s workflow should be different and use
> force == 1.  Don't the same considerations apply there to any code
> that needs to handle normal connection cleanup, or dead connection
> cleanup?
_______________________________________________
dev-tech-ldap mailing list
[email protected]
https://lists.mozilla.org/listinfo/dev-tech-ldap

Reply via email to