Dan Hecht 54194af6ef IMPALA-4631: don't use floating point operations for time unit conversions ...

This was leading to the PlanFragmentExecutor::Close() DCHECK because
with floating point we can have c * a + c * b > c * (a + b).  Also note
this is much more likely to happen when using the MONOTONIC_COARSE since
that will result in the nested scoped timers ending up starting/stopping
at exactly the same time.  Additionally, the new code is faster.

Change-Id: I7237f579b201f5bd3930f66e9c2c8d700c37ffeb
Reviewed-on: http://gerrit.cloudera.org:8080/5434
Reviewed-by: Dan Hecht <dhecht@cloudera.com>
Tested-by: Jim Apple <jbapple-impala@apache.org>

2016-12-15 22:37:47 +00:00

..

generated-sources

IMPALA-3552: Make incremental stats max serialized size configurable

2016-11-15 03:22:11 +00:00

src

IMPALA-4631: don't use floating point operations for time unit conversions

2016-12-15 22:37:47 +00:00

.gitignore

IMPALA-3253: Modify gen_build_version.sh to always output the right version

2016-07-05 13:37:25 -07:00

.impala.doxy

IMPALA-4110: Clean up issues found by Apache RAT.

2016-09-14 22:09:24 +00:00

CMakeLists.txt

IMPALA-3200: move bufferpool under runtime

2016-11-22 07:31:34 +00:00