Havoc Pennington hp at redhat.com
Wed Sep 6 12:59:08 PDT 2006

John (J5) Palmieri wrote:
> On Wed, 2006-09-06 at 12:14 -0400, Havoc Pennington wrote:
>> Alexander Larsson wrote:
>>> On johns request, here is a version of the builtin thread primitives
>>> that only adds the dbus_threads_init_default() call instead of changing
>>> any behaviour.
>> Can you guys post the conversation behind this? The patch as I read it 
>> does change the behavior/API (removes ability to set thread funcs)
> I read through the thread which I must have missed earlier and came to
> the conclusion that the current threading solution works so I don't
> really want to be pulling out code at this point in time. 

What code did the previous patch pull out? You mean the at_location 
stuff I guess?

i.e. the issue is whether to allow late thread init (at_location) or 
whether to auto-init threads on first use of the library?

I'd suggest going with whichever we feel is right; it is not really 
(easily) changeable post-1.0. I don't see a lot of "freeze violation" 
type of risk, certainly not enough to have a long-term wrong API. The 
at_location stuff was pretty new/untested relative to the older way anyhow.

That said, I don't really know which way I feel is right, so ;-)
it probably does not matter too much.


More information about the dbus mailing list