[wx-dev] reason for r47229

[Paul Cornett]

Robin Dunn wrote:
> Hi Paul,
> 
> Do you remember the reason for the r47229 change? [1]  The SVN log
> just says "process pending wx events before sending idle events" but
> that doesn't tell us why the change was needed. The reason I ask is
> because when I updated OSAF's version of wx their function test
> suite[2] started having failures in unexpected places.  We tracked
> down the problem to the different order of idle events, pending
> events, and pending deletes of TLWs conflicting with assumptions that
> the functional test suite had made about this behavior.  Of course
> the application itself seems to be able to handle the change just
> fine, so it's not a huge issue, but until we can fix those
> assumptions I'll need to revert that change in OSAF's wx so the
> functional tests can keep working.  So I'd like to understand why the
> change was made.

It was in response to a user complaint that they were getting an idle
event before a previously posted wxEvent was processed. Before the
change, the order of handling pending events and pending deletes
depended on what order the wxIdleEvents were sent in, which is highly
dependent on the internals of the wxEvent dispatching mechanism. I
thought it would be better for the ordering to be explicitly
deterministic.