Nandish, For a very basic code coverage implementation using cmake, I have created:
* JIRA: https://issues.apache.org/jira/browse/MADLIB-1138 * PR: https://github.com/apache/incubator-madlib/pull/151 This is a basic code coverage implementation runs on my mac using the following and a brew installed lcov module: Configured with: --prefix=/Library/Developer/CommandLineTools/usr --with-gxx-include-dir=/usr/include/c++/4.2.1 Apple LLVM version 8.1.0 (clang-802.0.42) Target: x86_64-apple-darwin16.6.0 Thread model: posix InstalledDir: /Library/Developer/CommandLineTools/usr/bin Regards, -=e On Fri, Jul 14, 2017 at 1:50 PM, Ed Espino <esp...@apache.org> wrote: > Sure NJ, > > I'll create both (Jira and PR). > > -=e > > On Fri, Jul 14, 2017 at 1:46 PM, Nandish Jayaram <njaya...@pivotal.io> > wrote: > >> Thank you Ed, this is great. We will create a JIRA to incorporate this >> (including documenting the steps to use it). Or, you can also create a PR. >> :) >> >> NJ >> >> On Fri, Jul 14, 2017 at 12:31 PM, Ed Espino <esp...@apache.org> wrote: >> >> > NJ, >> > >> > I spent about an hour getting it setup on my mac. This supported the >> > MLP PR review. I threw some info together below. >> > >> > Hope it helps, >> > -=e >> > >> > ------------------ >> > >> > Add lcov through brew (brings in lcov and genhtml utilities): >> > brew install lcov >> > >> > "Total HACK" to add code coverage support into the build for C/C++ >> > code. This is quick and dirty so I could get to the data I needed for >> > the PR review. >> > >> > diff --git a/CMakeLists.txt b/CMakeLists.txt >> > index b2172ef3..db80986c 100644 >> > --- a/CMakeLists.txt >> > +++ b/CMakeLists.txt >> > @@ -104,7 +104,7 @@ if(CMAKE_COMPILER_IS_GNUCXX) >> > set(CMAKE_INCLUDE_SYSTEM_FLAG_CXX "-isystem ") >> > endif(APPLE) >> > elseif(CMAKE_C_COMPILER_ID MATCHES "Clang") >> > - set(CMAKE_CXX_FLAGS "-stdlib=libstdc++") >> > + set(CMAKE_CXX_FLAGS "-stdlib=libstdc++ -fprofile-arcs >> > -ftest-coverage") >> > endif(CMAKE_COMPILER_IS_GNUCXX) >> > >> > # force a `m4_' prefix to all builtins >> > @@ -114,6 +114,8 @@ else() >> > set(M4_ARGUMENTS "--prefix-builtins") >> > endif() >> > >> > +set(CMAKE_C_FLAGS "-fprofile-arcs -ftest-coverage") >> > + >> > # Read and parse Version.yml file >> > file(READ "${MADLIB_VERSION_YML}" _MADLIB_VERSION_CONTENTS) >> > string(REGEX REPLACE "^.*version:[ \t]*([^\n]*)\n.*" "\\1" >> > MADLIB_VERSION_STRING "${_MADLIB_VERSION_CONTENTS}") >> > >> > Once MADlib is built and installed, run your tests: >> > >> > ## Build and install MADlib >> > mkdir build >> > cd build >> > cmake .. >> > make -j8 install >> > /usr/local/madlib/bin/madpack -s madlib -p postgres install >> > >> > ## At this point you can run your tests (I'm focusing on MLP) >> > >> > /usr/local/madlib/bin/madpack -s madlib -p postgres install-check -t >> > convex/mlp >> > >> > ## Time to capture the results and generate html report. >> > >> > lcov --capture --directory . --output-file coverage.info >> > genhtml coverage.info --output-directory gcov >> > >> > # You will notice there is some uninteresting system coverage info. >> > # The following will filter them out. >> > >> > lcov --remove coverage.info '/usr/include/*' '/usr/local/include/*' >> > '/usr/local/postgres/*' '*build/third_party/*' -o >> coverage_filtered.info >> > rm -rf gcov >> > genhtml coverage_filtered.info --output-directory gcov >> > >> > It is possible to zero out the counters with the following. >> > lcov --zerocounters --directory . >> > >> > >> > On Fri, Jul 14, 2017 at 12:23 PM, Nandish Jayaram <njaya...@pivotal.io> >> > wrote: >> > >> > > Hi Ed, >> > > >> > > We haven't set that up for MADlib yet, but we will be looking into it >> > soon. >> > > Any ideas? >> > > >> > > NJ >> > > >> > > On Fri, Jul 14, 2017 at 10:00 AM, Ed Espino <esp...@apache.org> >> wrote: >> > > >> > > > Out of curiosity, do MADlib developers regularly use code coverage >> > > > utilities to measure the coverage quality of their tests? >> > > > >> > > > -=e >> > > > >> > > > -- >> > > > *Ed Espino* >> > > > >> > > >> > >> > >> > >> > -- >> > *Ed Espino* >> > >> > > > > -- > *Ed Espino* > -- *Ed Espino*
