Re: When must slap_callback.sc_writewait be set?

2016-06-09 Thread Hallvard Breien Furuseth

On 08. juni 2016 23:55, Howard Chu wrote:

Hallvard Breien Furuseth wrote:

I lost this thread (from May 19).  Summarized in ITS#8435
plus a new suggestion: Revert sc_writewait and get it
done with a field somewhere else, maintained by callbacks.
Dunno if that's feasible.


There is no "somewhere else" to use.

I see no reason to revert the change.


The reason would be trying to not break third-party modules, like I've said.
But the problem is 1.5 years old and hopefully such a problem would have
been reported by now, so maybe there are very few affected modules out there.


We can simply add sc_writewait to the
end of the structure, which is the normal practice for adding new fields to
existing structs.


--
Hallvard



Re: When must slap_callback.sc_writewait be set?

2016-06-09 Thread Howard Chu

Hallvard Breien Furuseth wrote:

On 08. juni 2016 23:55, Howard Chu wrote:

Hallvard Breien Furuseth wrote:

I lost this thread (from May 19).  Summarized in ITS#8435
plus a new suggestion: Revert sc_writewait and get it
done with a field somewhere else, maintained by callbacks.
Dunno if that's feasible.


There is no "somewhere else" to use.

I see no reason to revert the change.


The reason would be trying to not break third-party modules, like I've said.
But the problem is 1.5 years old and hopefully such a problem would have
been reported by now, so maybe there are very few affected modules out there.


Exactly.



We can simply add sc_writewait to the
end of the structure, which is the normal practice for adding new fields to
existing structs.





--
  -- Howard Chu
  CTO, Symas Corp.   http://www.symas.com
  Director, Highland Sun http://highlandsun.com/hyc/
  Chief Architect, OpenLDAP  http://www.openldap.org/project/