I am attempting to use XcodeCoverage to get code coverage reports but am finding that lines that I know are being executed are not showing in the coverage reports or the raw gcov files. I followed the XcodeCoverage installation guide in multiple projects and am seeing the same results.
I created a sample project that exhibits the same behavior, what am I doing wrong here?
After trying many different things, the answer was to not include the main target's classes in the test target's "Target Membership".
Related
I created a multi-module project with a single sonar-project.properties. Most of the project is in c but one module is c++. I left the sonar.language property blank and ran sonar-runner with analysis.
The project and modules were created in the database and everything looked OK but when I did a preview on the same code I had a large number of new issues on the c++ project.
On closer inspection I found that the module in sonarqube was all wrong - it consisted of only the header files (.h) and most of the issues were the use of C style comments. It seems that the module was identified as a C module so all the .cpp files were ignored. However, in the preview the language was identified correctly and issues were found in the .cpp files that were clearly not know to sonarqube.
I cannot find any place in the UI to change the language of a module nor can I figure out a way to force it after the fact in my sonar-project.profiles.
First, sonar.language is apparently deprecated, and second, using module.sonar.language=cpp caused an error:
Caused by: sonar.profile was set to 'DEFAULT_C' but didn't match any profile for any language. Please check your configuration.
DEFAULT_C is something that we use for projects generally but not the one I'm working on. I have different profiles set up for different languages via the UI, but if I cannot get the analysis to see the correct language it doesn't help me much.
Any one have any idea how I can force the language of a module after the fact?
Hmm. I considered just deleting this but if I had found an answer on stackoverflow it would have eased my pain a little.
Brute force works.
Just set the language and the profile you want for the moudle and ignore complaints about such things being deprecated. Chosing a language won't force the right profile even if you have set up an explicit profile for the language in the UI - so just state it in the sonar-project.properties file.
In my case:
module.sonar.language=cpp
module.sonar.profile=MY_CPP_PROFILE
Throw sonar-runner at that and it will beat the backend into shape.
MSTest + Visual Studio generates accessors which appear on the NCover coverage report. They usually have a low coverage in my case. I wonder if I use them in my tests if the method I'm testing will appear as covered in the Accessor but not in the original source code.
If this is the case, it is a problem and I need a solution.
The dev team was interested in this issue and created an MSTest private accessor project in VS 2010.
I ran coverage on it in NCover 3 with MSTest, and it appears that these accessors are reflecting the source code, not a generated version of it, at least in our test.
If you'd like to get the test project and try it out, just contact us at support#ncover.com.
Thanks!
If I understand your scenario, then any method called by your test will be shown as covered in the source code.
If the test adds accessors, those will also be shown as covered where they appear, but you can filter them out by excluding the get and set methods that are generated.
Here's a link to the syntax for the method exclude:
http://docs.ncover.com/ref/3-0/ncover-console/command-line/profiling-options#em
We've been successful here with the regex .*.get_.*, just as an example.
NCover Support
I have code coverage in XCode 4.2 with libprofile_rt working. I have noticed that the coverage files are reset for every execution of my program. This is different from the way gcov worked and its something I need to work (since my tests execute across multiple runs of my app). Can someone point me to the way to make libprofile_rt generate cumulative coverage?
actually looks to be an existing bug in LLVM: http://llvm.org/bugs/show_bug.cgi?id=11457
You don't need to include *libprofile_rt.dylib* and use *-profile_rt* flag when running project in xCode 4.4.1 (works with iOS Simulator like a charm) Just set these flags:
Generate Test Coverage Files: YES
Instrument Program Flow: YES
Build and run. In case you runt Test-target of your app, *.gcda files are created automatically because tested app terminates, in case of running regular target, *.gcda files are created when you explicitly kill your app (the other approach is to set UIApplicationExitsOnSuspend = YES in .plist)
Please note, it was well tested on new project created with xCode 4.4.1
Code coverage files will be output to /Build/Intermediates/.build//.build/Objects-normal/
I have been trying to generate code coverage data for my application on XCode 4.2.1 (build 4D502) using the steps outlined at http://code.google.com/p/coverstory/wiki/UsingCoverstory
I am obviously missing something as there are no .gcno or .gcda files being generated along side my object files. I've tried adding the settings to both my test target and my application target but to no avail. I've had a second set of eyes go over things and all appears to be in order. Any ideas as to what is missing. Thanks in advance.
It doesn't work with llvm-gcc; you'll need to use the "LLVM Compiler 3.0" (aka clang). (I assume by the 'llvm-gcc' tag that you're not currently using it.)
I did this just last week, so it's definitely working.
I am using Xcode 3.2 on 10.6, with the shipped version of gcov and default GCC compiler (both version 4.2.1). I have created a dependent Cocoa unit test bundle which is injected into my app, and followed Apple's documentation on setting up a gcov-instrumented build configuration - based on the Debug configuration which doesn't have any compiler optimisations enabled.
When I build the test bundle with this 'Gcov-instrumented' configuration, the app launches and the tests are injected and run. Also the coverage statistics files are generated at:
build/<AppTarget>.build/Gcov-instrumented/<AppTarget>.build/Objects-normal/x86_64/<object>.gcda
build/<AppTarget>.build/Gcov-instrumented/<AppTarget>.build/Objects-normal/x86_64/<object>.gcno
So far so good. I know the tests are really being run because if I insert failures then the test suite fails as expected. Unfortunately, gcov reports that no lines of the objects have been covered by the tests! Every line is reported as 0 coverage. I've searched here and at the Apple mailing list archives, and can't find anyone with a similar problem. I expect I'm missing something - but what is it?
I have been running into this problem intermittently. I stumbled into a good length of lucky time where it was "mostly working" in my main project. By mostly working I mean I was able to get coverage to show up, but with a little more pain than on 10.5. I had to aggressively delete the coverage files and rebuild in order to get any updates, for instance. Even in my "barely working" state, the compiler/runtime doesn't seem willing to overwrite an existing coverage data file.
I think I've stumbled back on to a working scenario. Voodoo for sure, and less than ideal:
Make sure the SDK is 10.6.
Make sure the deployment is set to gcc 4.0 (!).
Do a full clean of the project.
Rebuild.
Does this get you the expected coverage data?
Granted, I would expect to be able to generate coverage files that work using gcc 4.2 or one of the LLVM compilers, but for the moment this seems to have me slightly less crippled than before.
Another clue: when I observe the coverage data (using Google's Cover Story), I get warnings like:
/BUILDRESULTS/MarsEdit.build/Code Coverage/MarsEditDataModelTests.build/Objects-normal/i386/MEDataItem.gcno:version '400A', prefer '402*'
But the coverage information is still presented. So it seems it's balking at an older coverage format version, but if I use the newer compiler (4.2) which would presumably generate the newer format .gcno files, the 0% coverage problem plagues me as it does Graham.
In Xcode 4.6, coverage reports just seem to work. Turn on "Generate Test Coverage Files" and "Instrument Program Flow" for the app target, run the tests, then point the coverage tool at the output. That's in ~/Library/Developer/Xcode/Derived Data/<Project>-<hash>/Build/Intermediates/<Target>.build/Debug-iphonesimulator/<Target>.build; opening that folder in CoverStory shows me the results I expected.
This GitHub repository contains scripts that can automatically discover the output and get lcov to visualise it; I'm happy with CoverStory for now.
You can get code coverage working with gcc 4.2.1. Details here:
SnowLeopardGCov
I have been trying to get the Code coverage working for iPhone simulator and always get a 0% coverage. Below are the configuration details and the steps that I have tried.
Configuration
Xcode 3.2.5/iOS 4.1 and iOS 4.2/Mac 10.6/GCC 4.2
Application
UICatalog
References
http://www.cubiclemuses.com/cm/articles/2009/05/14/coverstory-on-the-iphone/
http://developer.apple.com/library/mac/#qa/qa2007/qa1514.html
Steps
Enable “Generate Test Coverage Files”
Enable “Instrument Program Flow”
Add “-lgcov” to “Other Linker Flags”
UIApplicationExitsOnSuspend flag in Info.plist is set to true
Result
I have the .gcda files generated but the coverage always show 0%.
Settings tried
Changing GCC to 4.0 and 4.2. When I try to change the GCC to 4.0 I get 26 build errors.
Set environment variables:
const char *prefix = "GCOV_PREFIX";
const char *prefixValue = [[NSHomeDirectory() stringByAppendingPathComponent:#"Documents"] cStringUsingEncoding:NSASCIIStringEncoding]; // This gets the filepath to the app's Documents directory
const char *prefixStrip = "GCOV_PREFIX_STRIP";
const char *prefixStripValue = "1";
setenv(prefix, prefixValue, 1); // This sets an environment variable which tells gcov where to put the .gcda files.
setenv(prefixStrip, prefixStripValue, 1); // This tells gcov to strip the default prefix, and use the filepath that we just declared.)
GCC Optimization set to None (-O0) and unchecked the precompiled prefix header file flag.
I was getting 0 coverage when *.gcno files for static library was overwritten by those for shared library.