Re: [3/3] git commit: Adding a new feature to allow for shards to pull directly from a queue like interface.
On Sat, Feb 22, 2014 at 9:47 PM, <[EMAIL PROTECTED]> wrote:
Hmm... there's a timing issue here. Any subclasses that implement
queue-like things are likely to have costly initialization work to do.
The problem is that this thread starts [or could start] "taking"
before the subclass' constructor has completed its initialization. If
we keep this, then every implementation would be forced to create a
block on the initial take.
We could move the runnable to an inner class, but that'd force callers
to call 'start' or 'listen' or somesuch, which we could use to start
the thread... thoughts?