I am able to generate coverage.dat files with bazel command:
bazel coverage //tests/... --instrumented_filter=/src[/:]
This generates report for one of the classes, because coverage.dat files are generated separately for each instrumented file in different directories. How do I get a merged coverage.dat?
The coverage.dat report should contain coverage information about all the classes affected by the --instrumentation_filter. This file should be located under bazel-testlogs/path/to/your/package/TestTarget.
You shouldn't have to write anything additional. Bazel does generate multiple temporary .dat files, but it merges all of them in the final coverage.dat file, whose location is printed by bazel when it finishes to run. That file is the one with the location I described above. Make sure to check that file and check if you're using --instrumentation_filter (*) correctly.
(*) From the command line manual:
When coverage is enabled, only rules with names included by the
specified regex- based filter will be instrumented. Rules prefixed
with '-' are excluded instead. Note that only non-test rules are
instrumented unless -- instrument_test_targets is enabled.
Related
I am working on creating a Jenkins pipeline for unit-testing maybe with GTest.
My plan is to use the following tools:
GTest for Unit-Testing, gcov for generating gcda and gcno Files and gcovr for xml or Html outputs of the unit-testing results.
It's working well till now with the help from the internet and particularly stack overflow.
But I am struggling with 3 issues.
gcov is creating gcda and gcno files for gtest sources and my unit-tests. Because gcovr is invoking them and I see them in the HTML files. how can I avoid this? I only want my production code in the HTML files.
I can only see code coverage for template classes if gcov is generating gcda and gcno files for my unit-tests. So I need a simple idea for 1) Maybe can I use an exclude flag in gcovr?
Unused functions in template classes (inline functions) are not covered. Code coverage is always 100% but I tried with different flags, and nothing helped.
-fprofile-abs-path --coverage -fno-inline -fno-inline-small-functions -fno-default-inline -fkeep-inline-functions
I added a picture to show, what I am talking about. UnitTests and GTests covering results should not appear in gcovr HTML...
You can filter out unwanted coverage data, but you can't create data that doesn't exist.
1. gcov is creating gcda and gcno files for gtest sources and my unit-tests. Because gcovr is invoking them and I see them in the HTML files. how can I avoid this? I only want my production code in the HTML files.
Use gcovr --exclude GoogleTest/ --exclude UnitTests/
Gcovr has a per-file filtering system that allows you to specify which source code files to include/exclude. For a file to be included in the coverage report,
any --filter pattern must match, and
no --exclude pattern must match.
Or phrased in reverse: a file is excluded if it doesn't match any --filter or if it matches any --exclude pattern.
If you don't provide an explicit --filter, then the default filter is the --root directory, which in turn defaults to the current working directory.
These patterns are regexes. Usually, these are used to match paths relative to the current working directory. For example, you can limit the reports to a src/ directory with gcovr --filter src/. Or you can exclude the GoogleTest/ directory with gcovr --exclude GoogleTest/.
Gcovr also has a way to filter gcda/gcno files (search_paths and --gcov-filter), but that is mostly useful as a performance optimization.
2. I can only see code coverage for template classes if gcov is generating gcda and gcno files for my unit-tests. So I need a simple idea for 1) Maybe can I use an exclude flag in gcovr?
This is by design. As explained above, you can solve this via gcovr's exclude flag.
You get a gcda/gcno file per compilation unit. Header files are included into multiple compilation units, so their coverage information is essentially split across all compilation units that include it.
So, if you want coverage for code in header files, and you include these headers into your unit tests, then gcovr must also process the gcda/gcno files from those unit tests.
3. Unused functions in template classes (inline functions) are not covered. Code coverage is always 100% but I tried with different flags, and nothing helped. -fprofile-abs-path --coverage -fno-inline -fno-inline-small-functions -fno-default-inline -fkeep-inline-functions
The gcov coverage data model works on an assembly-code level. Counters are inserted by the compiler itself, but only for functions for which the compiler actually generates machine code. Thus, as far as gcov is concerned, inline functions, optimized-out code, and non-instantiated templates simply do not exist.
This is quite annoying, but it's potentially difficult to work around.
This can most clearly be avoided by making sure that all functions for which you want coverage data are referenced via your unit tests. It is not necessary to actually invoke that function, merely referencing it should be sufficient. For example, I'd write a function to ignore() arbitrary values despite optimizations, then:
ignore(&some_inline_function);
Possible implementation: template<class T> void ignore(T const& t) { volatile T sinkhole = t; }
Your suggested options like -fno-inline do not work because the code for these functions isn't generated in the first place.
With GCC and when using C++ (but not C), the -fkeep-inline-functions should work, but only for non-templated inline functions.
If a non-templated inline function is only used within one file and isn't provided in a header to multiple files, then it should instead be declared static (in C) or in an anonymous namespace (in C++11 or later), so that -Wunused-function or -Wall notify you if it isn't referenced.
Templates are more tricky in general. Each distinct instantiation of a template results in separate functions. Gcovr does aggregate coverage data across them, but in order for the template to appear in the coverage data it must be instantiated at least once. You will have to do this manually.
Given I have a list of files, e.g foo/src/main.cpp, foo/src/bar.cpp, foo/README.md is it possible to determine which of those files are part of a bazel package?
In my example, the output would e.g. be foo/src/main.cpp, foo/src/bar.cpp since the README.md would not be part of the build.
One way to do this would be to call bazel query on each file and see if it results in an output, but that is quite inefficient and so I was wondering if there is an easier way.
Background: I am trying to determine if a changes in a set of files have an impact on a target, and I want to use bazel query somepath(//some/target, set($FILES)) for that, but this will fail if any of the files in $FILES is not part of a BUILD file.
How about flipping it around and querying for all the source files of the target with:
bazel query 'kind("source file", deps(//some:target))'
and then checking if the result has any of the files in the set
please take a look at the bin-win target in my repository here:
https://github.com/thinlizzy/bazelexample/blob/master/demo/BUILD#L28
it seems to be properly packing the executable inside a file named bin-win.tar.gz, but I still have some questions:
1- in my machine, the file is being generated at this directory:
C:\Users\John\AppData\Local\Temp_bazel_John\aS4O8v3V\execroot__main__\bazel-out\x64_windows-fastbuild\bin\demo
which makes finding the tar.gz file a cumbersome task.
The question is how can I make my bin-win target to move the file from there to a "better location"? (perhaps defined by an environment variable or a cmd line parameter/flag)
2- how can I include more files with my executable? My actual use case is I want to supply data files and some DLLs together with the executable. Should I use a filegroup() rule and refer its name in the "srcs" attribute as well?
2a- for the DLLs, is there a way to make a filegroup() rule to interpret environment variables? (e.g: the directories of the DLLs)
Thanks!
Look for the bazel-bin and bazel-genfiles directories in your workspace. These are actually junctions (directory symlinks) that Bazel updates after every build. If you bazel build //:demo, you can access its output as bazel-bin\demo.
(a) You can also set TMP and TEMP in your environment to point to e.g. c:\tmp. Bazel will pick those up instead of C:\Users\John\AppData\Local\Temp, so the full path for the output directory (that bazel-bin points to) will be c:\tmp\aS4O8v3V\execroot\__main__\bazel-out\x64_windows-fastbuild\bin.
(b) Or you can pass the --output_user_root startup flag, e.g. bazel--output_user_root=c:\tmp build //:demo. That will have the same effect as (a).
There's currently no way to get rid of the _bazel_John\aS4O8v3V\execroot part of the path.
Yes, I think you need to put those files in pkg_tar.srcs. Whether you use a filegroup() rule is irrelevant; filegroup just lets you group files together, so you can refer to the group by name, which is useful when you need to refer to the same files in multiple rules.
2.a. I don't think so.
Problem
I wonder how to inform bazel about dependencies unknown at declaration time, but known at build time (a.k.a implicit dependencies, dynamic dependencies, ...). For instance when compiling C++ sources, the .cpp source file will depends on some header files and this information is not available when writing the BUILD file. It needs to be retrieve at build time. Whatever is the solution to get the information (dry-run, generating depfile, parsing stdout), it needs to be done at build time and the information need to be retrieved to bazel build graph.
Since skylark does not allow to do I/O, for instance to read a generated depfile or to parse stdout result containing a dependency list, I have no clue on how to deal with it.
Behind implicit dependencies, I am looking for correct incremental build.
Example
To experiment this problem I have created a simple tool, just_a_tool.exe which takes an input file, read a list of file from it, and concatenate the content of all these file to an output file.
command line example:
just_a_tool.exe --input input.txt --depfile dep.d output.txt
dep.d contains the list of all the read files.
Issue
If I change the content of test1.txt, test2.txt, or test3.txt, bazel does not rebuild output.txt file. Of course, because it does not know there were dependencies.
Example files
just_a_tool.bzl
def _impl(ctx):
exec_path = "C:/Code/JustATool/just_a_tool.exe"
for f in ctx.attr.source.files:
source_path = f.path
output_path = ctx.outputs.out.path
dep_file = ctx.actions.declare_file("dep.d")
args = ["--input", source_path, "--dep_file", dep_file.path, output_path]
ctx.actions.run(
outputs=[ctx.outputs.out, dep_file],
executable=exec_path,
inputs=ctx.attr.source.files,
arguments=args
)
jat_convert = rule(
implementation = _impl,
attrs = {
"source" : attr.label(mandatory=True, allow_files=True, single_file=True)
},
outputs = {"out": "%{name}.txt"}
)
BUILD
load("//tool:just_a_tool.bzl", "jat_convert")
jat_convert(
name="my_output",
source=":input.txt"
)
input.txt
test1.txt
test2.txt
test3.txt
Goal
I want to do correct and fast incremental build for the following situation:
Generate reflection data from C++ sources, this custom tool execution depends on header file included in my source files.
Use a internal tool to build asset file which can include other files
Run a custom preprocessor on my shaders allowing a #include feature
Thanks!
Bazel's extension language doesn't support creating actions with a dynamic set of inputs, where this set depends on the output of a previous action. In other words, custom rules cannot run an action, read the action's output, then create actions with those inputs or update (or prune the set of) inputs of already created actions.
Instead I suggest adding attribute(s) to your rule where the user can declare the set of files that the sources may include. I call this "the universe of headers". The actions you create depend on this user-defined universe, so the set of action inputs is completely defined. Of course this means these actions potentially depend on more files than the cpp files, which they process, include.
This approach is analogous to how the cc_* rules work: a file in cc_*.srcs can include other files in the srcs of the same rule and from hdrs of dependencies, but nothing else. Thus the union of srcs + hdrs of (direct & transitive) dependencies defines the universe of header files that a cpp file may include.
I've got an iOS framework that has a dependency on the (presumably Google maintained) pod called '!ProtoCompiler'. In order to build my framework I'm going to need it in the sandbox. So, I have a genrule and can try to include it with
src = glob(['Pods/!ProtoCompiler/**/*']) but I get the following error:
ERROR: BUILD:2:1: //Foo:framework-debug: invalid label 'Pods/!ProtoCompiler/google/protobuf/any.proto' in element 1118 of attribute 'srcs' in 'genrule' rule: invalid target name 'Pods/!ProtoCompiler/google/protobuf/any.proto': target names may not contain '!'.
As is, this seems like a total blocker for me using bazel to do this build. I don't have the ability to rename the pod directory as far as I can tell. As far as I can tell, the ! prohibition is supposed to be for target labels, is there any way I can specify that this is just a file, not a label? Or are those two concepts completely melded in bazel?
(Also, if I get this to work I'm worried about the fact that this produces a .framework directory and it seems like rules are expected to produces files only. Maybe I'll zip it up and then unzip it as part of the build of the test harness.)
As far as I can tell, the ! prohibition is supposed to be for target
labels, is there any way I can specify that this is just a file, not a
label? Or are those two concepts completely melded in bazel?
They are mostly molded.
Bazel associates a label with all source files in a package that appear in BUILD files, so that you can write srcs=["foo.cc", "//bar:baz.cc"] in a build rule and it'll work regardless of foo.cc and baz.cc being a source file, a generated file, or a build rule's name that produces files suitable for this particular srcs attribute.
That said you can of course have any file in the package, but if the name won't allow Bazel to derive a label from it, then you can't reference them in the BUILD file. Since glob is evaluated during loading and is expanded to a list of labels, using glob won't work around this limitation.
(...) it seems like rules are expected to produces files only. Maybe
I'll zip it up and then unzip it as part of the build of the test
harness.
Yes, that's the usual approach.