RFC: rr release 0.1
jones.chris.g at gmail.com
Tue May 28 15:09:33 PDT 2013
I think we should release early and often, because (i) good advertising;
(ii) relatedly, helps attract contributors; (iii) gets useful features into
users' hands as soon as possible. Of course, part (iii) is the tricky
part, especially for a first release: what's the threshold of usefulness we
need to cross before releasing?
I think that once we get basic gdb support finished up, we'll have a tool
with a compelling use case: deterministic debugging, even if only in the
forward direction (and for some workloads, faster-than-real-time replay).
Reverse replay can be simulated with binary search in the forwards
direction. We do know that rr's scheduling can be improved (issue #40),
but I don't think we need to resolve that for a 0.1 release.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the rr-dev