On Tue, Jun 16, 2020 at 10:40 AM Joe Orton <[email protected]> wrote: > > On Thu, May 07, 2020 at 09:50:26AM -0400, Eric Covener wrote: > > On Thu, May 7, 2020 at 9:39 AM Joe Orton <[email protected]> wrote: > > > Better question... or stupider question? For "modern" OpenLDAP where > > > ldap_set_rebind_proc takes a void *, this linked list cache is > > > completely redundant and you can "simply" pass the (bindDN, bindPW) > > > through to the rebind callback via the void *, and that will work > > > correctly? > > > > That makes sense, I would think so. The manual on my Mac has the > > userdata parm too. > > I have merged that change in r1878890 - thanks a lot for reviewing & > walking me through this far, apologies for moving slowing but I wanted > to get a working test case before proceeding. > > Reviewing this thread again I am not sure if the change to > remove/reorder the removal of the rebind is safe for non-OpenLDAP > toolkits. I can revert that bit if required? > > https://github.com/apache/httpd/commit/130eac3ae6e8f7a8465c7792660ce8b747642b23#diff-f0cf47f5d582509d36e95f170231bc21L226
I think it should be left in. I found I had made a very similar change in the distribution I look after while adding Darwin sandbox support.
