I'm trying to patch PROTOBUF_FINAL out of generated protobuf files, in order to do this I've created a simple bash script that will do so for me, but the problem I'm now facing is that I haven't been able to get it to run after protobuf_generate itself, only before it or not at all.
Internally it seems protobuf_generate runs add_custom_command for each file provided to it, and after all of that it sets an output variable containing all the generated files (_generated_srcs_all), so I tried depending on that with a custom command but with that it just never runs.
Did you try to depend on the generated files list which is stored in this variable ?
note: supposing you are talking about the function protobuf_generate
https://github.com/protocolbuffers/protobuf/blob/ee35b2da4b6d633eadcc105e5232319b47b494a7/cmake/protobuf-config.cmake.in#L141
Related
I'm diagnosing a Visual Studio project building problem, and I want to see what MSBuild.exe actually does. However, from Procmon, I see MSBuild creates BAT file, call it and delete it immediately, then how can I see that BAT's content?
For example, real BAT command is hidden inside tmpd60d571fd9d549e5b2b31bb1f2ba51a7.exec.cmd.
"C:\Windows\system32\cmd.exe" /Q /C C:\Users\win7evn\AppData\Local\Temp\tmpd60d571fd9d549e5b2b31bb1f2ba51a7.exec.cmd
MSBuild.exe writes cl.exe, link.exe, Prebuild/Postbuild custom commands etc into those temporary bat files. Specifically, I want to see how exactly MSBuild is calling my custom commands and how does he pass parameters to me from within a bat file.
The ideal way, I think, may be some tools that can intercept file deleting operation, and make a backup(copy) of that file, so we can investigate the "deleted" files later. Procmon only records ReadFile and WriteFile byte count, but not their byte content, so I think Procmon is not enough.
==== Update ====
According to stijn's answer(env-var MSBUILDPRESERVETOOLTEMPFILES=1), I tried it on VS2019 with success. The content in the xxx.exec.cmd is like this:
Two things to note:
We don't need to pass /verbosity:d parameter.
On VS2010, the MSBuild version may be too old, no effect.
Final comment: I think this feature of "seeing true .bat content" is critical, because, I may use many macros in my Custom-build/Prebuild/Postbuild commands, like $(ProjectDir), $(Configuration), or even my own defined macros of any name, so I need a global view of the macro expansion result, instead of checking each macro from VSIDE UI one by one(so many mouse clicking and time consuming) . The .bat content is the exact global view I need.
Msbuid creates .cmd files when it runs the Exec task, depending on what problem you are diagnosing it might be enough to just know what statements are added in those files, which can be found here: https://github.com/dotnet/msbuild/blob/main/src/Tasks/Exec.cs.
Otherwise set the MSBUILDPRESERVETOOLTEMPFILES environment variable to 1 so the files do not get deleted, then run with /v:D to see which files were created.
I'm using Karate in this way; during application execution, I get the test files from another source and I create feature files based on what I get.
then I iterate over the list of the tests and execute them.
My problem is that by using
CucumberRunner.parallel(getClass(), 5, resultDirectory);
I execute all the tests at every iteration, which causes tests to be executed multiple times.
Is there a way to execute one test at a time during application execution (I'am fully aware of the empty test class with annotation to specify one class but that doesn't seem to serve me here)
I thought about creating every feature file in a new folder so that I can specify the path of the folder that contains only one feature at a time, but CucumberRunner.parallel() accepts Class and not path.
Do you have any suggestions please?
You can explicitly set a single file (or even directory path) to run via the annotation:
#CucumberOptions(features = "classpath:animals/cats/cats-post.feature")
I think you already are aware of the Java API which can take one file at a time, but you won't get reports.
Well you can try this, set a System property cucumber.options with the value classpath:animals/cats/cats-post.feature and see if that works. If you add tags (search doc) each iteration can use a different tag and that would give you the behavior you need.
Just got an interesting idea, why don't you generate a single feature, and in that feature you make calls to all the generated feature files.
Also how about you programmatically delete (or move) the files after you are done with each iteration.
If all the above fails, I would try to replicate some of this code: https://github.com/intuit/karate/blob/master/karate-junit4/src/main/java/com/intuit/karate/junit4/Karate.java
I want to use a property ('currentId') which has a certain start value. For each test case the value should be increased by 1. I can do that by adding an extra test step in each test case which increases the value but that would be much copy paste. The code for that would be (see reference):
def uniqueUserPortion = testRunner.testCase.testSuite.project.getPropertyValue("currentId")
// convert it to an Integer, and increment
def uniqueUserPortionInc = uniqueUserPortion.toInteger() + 1
// set the property back as string
testRunner.testCase.testSuite.project.setPropertyValue("currentId", uniqueUserPortionInc.toString())
To avoid that copy&paste I've added the code above to the Load Script of the project but it doesn't work:
testSuite.testCases.each {
*code above*
}
What can I do to use the code in one script/call for all test cases?
I could define the property as the start value plus the test case ID but that would be a definition in each test case again since I can not reference the #TestCase#ID in project level/property.
Issue with what your are trying
Load Script of the project is executed once when you import the project into soapui workspace. So, this approach does not work.
As you rightly mentioned, either you need to have it in a separate step of the each test case or you can add the same code as setup script. Yes, it is copy paste only
It is possible to achieve easily using SoapUI NG which pro edition using Event feature.
Then your next question may be : how to do it in Open Source edition of SoapUI.
Here is an soapuiExtensions which I did sometime ago which allows you do the same without having to copy paste for each test case in open source edition.
All you need do is have your groovy script into a specific file called 'TestCaseBeforeRun.groovy'. That means, the script is executed before running each test case.
For more details refer README
This soapuiExtensions library allows users to have some additional functionality in soapUI(free edition) tool, like soapui pro allows to do something before, after doing something.
For eg: User may want to do something before running a test case or after running a test case etc by implementing appropriate groovy script as required. Allow me to add an example here. Usually user may want to add credentials for the request step automatically, see the script samples/scripts/TestSuiteTestStepAdded.groovy
How to use this library:
set SOAPUI_HOME environment variable.
copy lib/SoapUIExtListeners.jar file under $SOAPUI_HOME/bin/ext directory
copy samples/listeners/custom-listeners.xml file under $SOAPUI_HOME/bin/listeners directory
copy samples/scripts directory under $SOAPUI_HOME
And implement appropriate groovy script available under $SOAPUI_HOME/scripts. Refer Mappings file in order to implement respective groovy script.
Note: for windows users, you may need to check %SOAPUI_HOME%\bin\soapui.bat which actually overwrites SOAPUI_HOME, need to fix soapui.bat script if requires.
Uses jdk 7, soapUI 4.5.1, and groovy 1.8.9
Dependency
log4j
UPDATE: this is realted to the note in the above.
As it was mentioned in the note, soapui.bat overrides SOAPUI_HOME environment variable on windows, needs to be tweaked a bit. May be you want to copy that groovy file under %SOAPUI_HOME%\bin\scripts (this is without tweaking soapui.bat)and retry. If your machine is linux then it should work if you copy the groovy file under $SOAPUI_HOME/scripts directory
is there any plug-in or other possibility to set an environment variable in CC.NET 1.4.2 to some generated value. I would like to pass to MSBuild some random value (can be a time stamp where to put some build reports). Afterwords all the generated report files from the randomly named dir will be merged to cc.net report.
The problem here is that I can't use the CCNetBuildDate + CCNetBuildTime environment variables, due to the format of CCNetBuildTime (HH:mm:ss), because : is not a valid character for directory name. I could use them if CC.NET supports ':' replacement by some other char (e.g. '-').
I can use MSBuild community task to create the output directory with the help of <Time>-task, the problem is that I don't know how to return to CCNet in which random dir the reports were produced.
I can't use the labeller either, because we have rewritten the labeller and it always returns the dummy label (I know that is very bad and changes ccnet logic, but currently I have no choice).
I can write a plug-in, but I would like to use as much default technologies as possible.
Many thanks,
Ovanes
Can't you just produce the report files in the normal project working directory and merge them from there? Every other external reporting tool works this way.
We have a standard MSBuild project file that is used for our different deployment stages (pre-stage, stage, live, etc). Since each deployment stage is performed on a different server we introduced a server parameter called $SourceDatabaseServer and used this extensively in each of the targets inside the project file. Note: This database server name could be different from the server name on which the build is run.
To assist us with the customization of this parameter, we created a response file for each deployment stage and subsequently defined a value for this parameter in the response file, e.g. /p:SourceDatabaseServer=SRC_DB_NAME.
This worked fine, until we created a new deployment stage in which this value had to be the current computer name. So we thought by using the $(COMPUTERNAME) reserved property in the response file (/p:SourceDatabaseServer=$(COMPUTERNAME)), this would do the trick, but it seems like this value is interpreted literally by MSBuild, and we consequently get an error that server $(ComputerName) could not be found.
Interestingly, when the $(COMPUTERNAME) property is used directly in the proj file it works, but as stated above, we do not necessarily want to use the computer name in all the cases.
Is there a way to still use the $(COMPUTERNAME) property in the response file and get MSBuild to interpret this correctly?
What if you use %COMPUTERNAME%?
$(VAR) is the syntax for variable expansion when you're "inside" the MSBuild system, but coming from the outside, I believe you'd have to use the shell environment variable expansion syntax, %VAR%.