-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/20018/#review41077
-----------------------------------------------------------

Ship it!


Alright, I'll get this committed! I made a number of comments for cleanup below 
but I didn't open them as "issues" since I've included the fixes in the commit.

Please let me know if anything was missed or if we should follow up with 
anything!


3rdparty/libprocess/src/metrics/metrics.cpp
<https://reviews.apache.org/r/20018/#comment74461>

    Let's remove this since 'context' no longer exists.



3rdparty/libprocess/src/tests/metrics_tests.cpp
<https://reviews.apache.org/r/20018/#comment74463>

    Please no using clause for methods like this since just seeing get() out of 
context is a bit unintuitive.



3rdparty/libprocess/src/tests/metrics_tests.cpp
<https://reviews.apache.org/r/20018/#comment74484>

    You don't need EXPECT_FLOAT_EQ for integral double values, right?



3rdparty/libprocess/src/tests/metrics_tests.cpp
<https://reviews.apache.org/r/20018/#comment74478>

    Looks like we could simplify this test to only do what is necessary:
    
    // Ensure JSON is empty.
    
    // Add a counter and gauge.
    
    // Ensure they are present in the JSON.
    
    // Remove them.
    
    // Ensure they are not present in the JSON.



3rdparty/libprocess/src/tests/metrics_tests.cpp
<https://reviews.apache.org/r/20018/#comment74475>

    Please favor names like "counter" and "gauge" over "c" and "g".



3rdparty/libprocess/src/tests/metrics_tests.cpp
<https://reviews.apache.org/r/20018/#comment74477>

    It seems like all of our tests in libprocess would need this, no?



3rdparty/libprocess/src/tests/metrics_tests.cpp
<https://reviews.apache.org/r/20018/#comment74476>

    Testing this seems unnecessary, let's opt to keep the test simple instead!



3rdparty/libprocess/src/tests/metrics_tests.cpp
<https://reviews.apache.org/r/20018/#comment74471>

    s/c/counter/



3rdparty/libprocess/src/tests/metrics_tests.cpp
<https://reviews.apache.org/r/20018/#comment74465>

    You can remove the need for this call to advance outside the loop by 
advancing _before_ the counter pre-increment in the loop.



3rdparty/libprocess/src/tests/metrics_tests.cpp
<https://reviews.apache.org/r/20018/#comment74470>

    Advancing the clock after incrementing does not achieve anything, so does 
it makes sense to only increment before?



3rdparty/libprocess/src/tests/metrics_tests.cpp
<https://reviews.apache.org/r/20018/#comment74469>

    Let's move the response down to below the expected object.



3rdparty/libprocess/src/tests/metrics_tests.cpp
<https://reviews.apache.org/r/20018/#comment74468>

    Let's add some newlines here to make it easier to read.


- Ben Mahler


On April 22, 2014, 7:56 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, 7:56 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/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