-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/20018/#review41044
-----------------------------------------------------------


Almost there!


3rdparty/libprocess/include/process/statistics.hpp
<https://reviews.apache.org/r/20018/#comment74406>

    Looking around at other libraries, "p999" and "p9999" seem to be pretty 
standard so I take back that this might confuse people. Let's revert, sound 
good?



3rdparty/libprocess/include/process/statistics.hpp
<https://reviews.apache.org/r/20018/#comment74405>

    Looks like there is a regression here! What happens on line 76 when 
position is negative? Likely the CHECK_LT on line 86 will fail!
    
    This seems more difficult to understand and this change introduces a bug, 
let's revert!



3rdparty/libprocess/src/tests/metrics_tests.cpp
<https://reviews.apache.org/r/20018/#comment74412>

    Is this rebased off master? This test looks to be already committed.



3rdparty/libprocess/src/tests/metrics_tests.cpp
<https://reviews.apache.org/r/20018/#comment74411>

    size_t?



3rdparty/libprocess/src/tests/metrics_tests.cpp
<https://reviews.apache.org/r/20018/#comment74413>

    Hm, we probably should check the other percentiles here as well, no?



3rdparty/libprocess/src/tests/metrics_tests.cpp
<https://reviews.apache.org/r/20018/#comment74414>

    This test is pretty difficult to digest, can you add some comments above 
each logical block describing what you're testing?
    
    s/JSONAddAndRemove/Snapshot/ ?



3rdparty/libprocess/src/tests/metrics_tests.cpp
<https://reviews.apache.org/r/20018/#comment74415>

    s/jsonResponse/expected/?



3rdparty/libprocess/src/tests/metrics_tests.cpp
<https://reviews.apache.org/r/20018/#comment74417>

    Is there any reason to overlap the removal of gauge and counter? Could we 
remove the gauge first to make this a bit easier to follow?



3rdparty/libprocess/src/tests/metrics_tests.cpp
<https://reviews.apache.org/r/20018/#comment74409>

    Most of our tests in libprocess (a few exceptions) omit the "Test" suffix 
here, can you s/MetricsTest/Metrics/?
    
    s/JSONStatistics/SnapshotSatistics/ ?



3rdparty/libprocess/src/tests/metrics_tests.cpp
<https://reviews.apache.org/r/20018/#comment74410>

    Increment consistently instead of pre _and_ post here?
    
    Use size_t?



3rdparty/libprocess/src/tests/metrics_tests.cpp
<https://reviews.apache.org/r/20018/#comment74408>

    Can we come up with an interval that generates more intuitive percentiles? 
One that a reader can intuit?



3rdparty/libprocess/src/tests/statistics_tests.cpp
<https://reviews.apache.org/r/20018/#comment74407>

    Thanks! [-5,5] seems much cleaner than the extra 0 and [-5,4]!


- Ben Mahler


On April 22, 2014, 6:20 p.m., Dominic Hamon wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/20018/
> -----------------------------------------------------------
> 
> (Updated April 22, 2014, 6:20 p.m.)
> 
> 
> Review request for mesos, Benjamin Hindman, Ben Mahler, and Vinod Kone.
> 
> 
> Bugs: MESOS-1036
>     https://issues.apache.org/jira/browse/MESOS-1036
> 
> 
> Repository: mesos-git
> 
> 
> Description
> -------
> 
> see summary.
> 
> 
> Diffs
> -----
> 
>   3rdparty/libprocess/include/process/metrics/metric.hpp 
> 6a384ded8a4b57fd6aee819d0337773018c45669 
>   3rdparty/libprocess/include/process/metrics/metrics.hpp 
> c20bb639e8ef79de63f0d0d56c2ea40a15a1f995 
>   3rdparty/libprocess/include/process/statistics.hpp 
> d046bffa00fa839909112f65d19dfb9af46ad2b3 
>   3rdparty/libprocess/src/metrics/metrics.cpp 
> 391295aea91e837bb856a40ef51d1c33d44371d8 
>   3rdparty/libprocess/src/tests/metrics_tests.cpp 
> abe1588c931b45a09294812974788aa74de44dd4 
>   3rdparty/libprocess/src/tests/statistics_tests.cpp 
> 478453fd60056603cf2eb96e56ac2df7e47a3e99 
> 
> Diff: https://reviews.apache.org/r/20018/diff/
> 
> 
> Testing
> -------
> 
> make check
> 
> 
> Thanks,
> 
> Dominic Hamon
> 
>

Reply via email to