Different git branches could be running on different xUnit versions and the current approach of choosing the latest from nuget doesn't work. xUnit can have breaking changes. Hence it should be possible to specify a local xUnit executable path. This is similar to #96 but I do not want to fetch xUnit from nuget, as that could slow down the build as well.

Note: we have xUnit checked in to source control.

When I attempt to tag a build using MRPP_AddBuildTags with multiple tags separated by a space, the build ends up with 1 tag and the space is replaced with </tag><tag>.

For example, I attempt to tag with this: env=sandbox deploy=Everything
And I end up with this single tag: env=sandbox</tag><tag>deploy=Everything

One of features in dotcover is _generate a report _with result after analysis in different formats (HTML|JSON|XML|NDependXML) .
More info: JetBrains DotCover Page

SonarQube accept import a results from the report generated from dotcover in HTML format.

In the meta-runner xUnit.net-dotCover is missing how enable for generate a report.

One option is add and extra step after execute metarunner.
This extra step execute in a poweshell or command line the following:

${path}/dotcover report /Source=xunitcoverage.dcvr /Output=ResultDotCover.html /ReportType=HTML
to generate this report.

This execution in sometimes fails because the xunitcoverage.dcvr not exists, I think when publish the artefact was deleted from the working folder.

I think if you execute dotcover report step how a part of the metarunner works fine.

Can you add this feature in the metarunner?

Now it happens so: the build is triggered, it picks up dependencies while queuing, and afterwards it may run only with that dependency versions which are linked while queuing. So, when the last configuration of the long queue starts, it still runs with yesterday's build.

If I put getting dependency as a step it will help taking the very last build.

Related to http://youtrack.jetbrains.com/issue/TW-33828

When building a project and Chokolatey is not installed on the build agent, the Chokolatey install script apparently defines a url parameter containing the value https://chocolatey.org/api/v2/package/chocolatey/ which gets passed along to gitversion.exe. Using a local git repository (no url is specified), the following command was executed by the metarunner: C:\ProgramData\Chocolatey\bin\gitversion.exe "C:\TeamCity\buildAgent\work\4d3ca3d2ab478a4a" /url 'https://chocolatey.org/api/v2/package/chocolatey/' /output 'buildserver' /UpdateAssemblyInfo true. Running the build again works as expected, since the install script is not executed.

For some reason PHPUnit, using the meta-runner-power-pack, doesn't generate a logfile. I have another project in the same Teamcity which doesn't have this problem. I've gisted the logfile(https://gist.github.com/zbrag/d86f587264380118f676), which does have some errors to report, so why doesn't phpunit/teamcity see those?

All I'm getting right now is an 'errorcode 1'

Metarunner for PHP Composer Install downloads composer.phar every time. While this is good (since who knows how old composer binary was installed in the system) the method used is dangerous and not recommended.

Current method simply downloads PHAR archive and simply executes it without checking of the signature. Proper method to programmatically download composer is to use simple bash script provided by authors.

When configuring PHPUnit to generate code coverage reports and if configuring failure conditions to fail on anything except number of tests TeamCity throws an error that says : 'Cannot calculate build metric 'number of covered lines' - no value'

The current approach to using Invoke-Expression to launch nuget, xunit and dotCover exe's will fail if there is a space in the working folder path. It can be fixed by using call operator "&" and single quoted paths in conjunction with Invoke-Expression. Examples below with original code commented out followed by fixed code:

NuGet:

#Invoke-Expression "$nugetExe install xunit.runner.console -source $xUnitNuget"
Invoke-Expression "& '$nugetExe' install xunit.runner.console -source $xUnitNuget"

xUnit:

#$xunitOnlyCommand = "$xunit $xunitArg"
$xunitOnlyCommand = "& '$xunit' $xunitArg"

dotCover:

#$command = "$dotCover $dotCoverArgs"
$command = "& '$dotCover' $dotCoverArgs"

Hi there, I was testing those Meta Runners which technically may help me quite a lot, but I need to specify the PHPUnit version because the latest one will work just with PHP 7 while I need PHP-Unit 5 to work with PHP 5.6

Apparently there is no way where I can specify which version I can use or a specific path to an existing phpunit binary of my choice.

Can you please point me to the right direction?

Thanks
Best regards
Andrea

The issue is xUnit.net + dotCover plugin is not working correctly with .net core 2, here is our settings for this build step:

Here is the error /warning message from Teamcity

Any Thoughts on that?

Thanks
Shady

When building my project I get

PowerShell Executable: C:\Windows\SysWOW64\WindowsPowerShell\v1.0\powershell.exe
[08:50:26] Working directory: C:\Users\teamcity\teamcityconf\BuildAgent\trick\work\a340315af7a6407e
[08:50:28] Chocolatey already installed
[08:50:28] GitVersion not installed; installing GitVersion
[08:50:28] ##teamcity[buildStatus text='The term 'C:\ProgramData\chocolatey\chocolateyInstall\chocolatey.cmd' is not recognized as the name of a cmdlet, function, script file, or operable program. Check the spelling of the name, or if a path was included, verify that the path is correct and try again.' status='FAILURE']

however with the latest version of chocolatey this directory does not exist

Using Teamcity 2017.2 i get :

.NET Coverage: Failed to find .NET coverage tool at dotCover.exe

[10:16:33][Step 16/22] Step 2/2: Run xUnit.net tests (.NET Process Runner) (1s)
[10:16:33][Step 2/2] Starting: C:\TeamCity\buildAgent\plugins\dotnetPlugin\bin\JetBrains.BuildServer.NUnitLauncher.exe #TeamCityImplicit
[10:16:33][Step 2/2] in directory: C:\TeamCity\buildAgent\work\3915e306ed37130f
[10:16:34][Step 2/2] error: file not found: MSIL
[10:16:34][Step 2/2] .NET Coverage: Failed to find .NET coverage tool at dotCover.exe
[10:16:34][Step 2/2] .NET Coverage: Failed to find .NET coverage tool at dotCover.exe
[10:16:34][Step 2/2] Process exited with code 0
[10:16:34][Step 2/2] Step Run xUnit.net tests (.NET Process Runner) failed

I found an old version (in another teamcity install made many years ago) that works, but i could not relate to any history in this git repository :

MR_XUnitNet_very_old.xml.txt

At the top of the ant script

<property name="phpunit.result.error" value="false" />

is set.

Properties are immutable.

The later defined conditions can't change the value of the property. So, if we
get an exit code > 0 from PHPUnit, "runTest" will never fail.

<target name="runTests" depends="getPhpUnit,runPhpUnitPhar,runPhpUnitRuntime,collectCoverage">
    <fail message="Unit test error." if="${phpunit.result.error}" />
</target>

Removing ths line

<property name="phpunit.result.error" value="false" />

solves the problem for me.

Given: working folder with solution project folders and files, and Nuget.Config file
When: running xUnit+dotCover step
AND restoring package 'xunit.runner.console.2.1.0'
Then: package will be restored at Nuget.Config specified folder, not working folder
AND dotCover console throws error: Target executable is not specified

When an xunit test fails the step is still reported as failed ( ie in red ) but subsequent tasks still execute even though the execute step are all set to "If all previous steps finished successfully"

The xUnit.net console running is returning exit code '1'

TeamCity server version is 9.1.6 (build 37459)

[Step 1/1] dotCoverArgs: cover coverage_settings.xml /LogFile=dotCoverXunitLog.txt /ReturnTargetExitCode
[Step 1/1] JetBrains dotCover Console Runner 10.0.2. Build 104.0.20151218.125453
[Step 1/1] xUnit.net Console Runner (64-bit .NET 4.0.30319.42000)

...

[Step 1/1] [JetBrains dotCover] Analysed application exited with code '1'

In regards to the semver-buildnumber meta-runner, would it be possible to support pre-release version numbers such as the following?

1.0.0-alpha1

Currently it only supports .NET version numbers that exactly conform to major.minor.build.release.

I'm trying to create a build step for test dll that tests an asp.net 5 web api project. The tests run just fine locally using xUnit, but in Team City, I get the following (I added ):

System.InvalidOperationException: Unknown test framework: could not find xunit.dll (v1) or xunit.execution.*.dll (v2) in D:\TeamCity\buildAgent\work\9e2adc6bf9610ed8\artifacts\bin\Release\dnx451

I've tried various combinations of NuGet package restore and dnx updates, but always get the same result. I've found that various people have had a similar issue, but all their solutions do not seem to work for me. Can someone lend a hand with this?

Setting a custom working directory doesn't let Composer search for the composer.json file in that directory.

There seems to be a lot of malformed XML in the version 3 file which TeamCity can't import.

GitVersion metarunner seems to be unable to persist "Update TeamCity build version" ("mr.GitVersion.output").

Steps to reproduce:

1. Create new step: GitVersion (flag is set by default).
2. Save.
3. Open step. Notice that Flag is now cleared.

TeamCity Professional 10.0 (build 42002).

I am install GitVersion runner, add build step with GitVersion, but 'AssemblyInfo patcher' build feature cannot update file with %system.GitVersion.AssemblySemVer% property.

PS. I use GitVersion 3.0.0

I'm using teamcity 2017.2 and it does not show up with

Export dotCover report

enabled
UPDATE:
i just noticed this error :
No available .NET Coverage report generator for type 'dotcover'. Skipped files: [C:\TeamCity\buildAgent\work\3915e306ed37130f\xunitcoverage.dcvr]
any clue ?

So, the param $nugetExe value should be
[string] $nugetExe = "%teamcity.tool.NuGet.CommandLine.DEFAULT%\tools\nuget.exe"
instead of
[string] $nugetExe = "%teamcity.tool.NuGet.CommandLine.DEFAULT.nupkg%\tools\nuget.exe"

Error message is as follows. I'm using MR_GitVersion3.xml.

The term 'Join-Path' is not recognized as the name of a cmdlet,  function, script file, or operable program. Check the spelling of the name, or if a path was included, verify that the path is correct and try again.

Seemingly relative line:

[string] $workingDir = (Join-Path "%teamcity.build.workingDir%" "%mr.GitVersion.gitCheckoutDir%"),

I'm adding a new agent to teamcity, it first download chocolatey, and then tries do install xunit, but it fails with the following error messages. Looking at choco folder it looks like the structure it changed in new versions before there was a C:\ProgramData\Chocolatey\chocolateyInstall subfolder, now there is not.
I think that the plugin should use choco in path rather than making any assumption on any installed folder.

Step 5/9: Test (xUnit.net + dotCover)
[12:37:32][Step 5/9] Step 1/2: Install xUnit from Chocolatey (Powershell)
[12:37:32][Step 1/2] PowerShell Executable: C:\Windows\SysWOW64\WindowsPowerShell\v1.0\powershell.exe
[12:37:32][Step 1/2] Working directory: C:\BuildAgent\work\ca594cc2eae5296
[12:37:32][Step 1/2] PowerShell arguments: [-NoProfile, -NonInteractive, -ExecutionPolicy, ByPass, -File, C:\BuildAgent\temp\buildTmp\powershell6473424193213794350.ps1]
[12:37:33][Step 1/2] Chocolatey already installed
[12:37:33][Step 1/2] xUnit.net not installed; installing xUnit.net
[12:37:33][Step 1/2] ##teamcity[buildStatus text='The term 'C:\ProgramData\Chocolatey\chocolateyInstall\chocolatey.cmd' is not recognized as the name of a cmdlet, function, script file, or operable program. Check the spelling of the name, or if a path was included, verify that the path is correct and try again.' status='FAILURE']
[12:37:33]
[Step 1/2] Incorrect property name.
Valid property list format is (name( )=( )'escaped_value'( )) where escape symbol is "|"
[12:37:33][Step 1/2] ##teamcity[message text='The term 'C:\ProgramData\Chocolatey\chocolateyInstall\chocolatey.cmd' is not recognized as the name of a cmdlet, function, script file, or operable program. Check the spelling of the name, or if a path was included, verify that the path is correct and try again.' status='ERROR']
[12:37:33]
[Step 1/2] Incorrect property name.
Valid property list format is (name( )=( )'escaped_value'( )) where escape symbol is "|"
[12:37:33][Step 1/2] Process exited with code 1
[12:37:33][Step 1/2] Step Install xUnit from Chocolatey (Powershell) failed

using

src\**\bin\Release\*.Tests.dll

Causes an exception as the gci call returns an Object[] and not a string as expected.

Wrapping the result in a [String]::Join fixes this, however under certain conditions a string is returned (e.g. explicit paths laid) so this would need to be defensively coded around.

We are using the version from July 31 in TC 2018.1.1. We specified particular credentials in the "Add Build Tags" build step. It works 99% of the time, but on rare occasions in random builds we get this:

[15:02:23]prepare-internal-credentials:
[15:02:23]
[15:02:23]prepare-custom-credentials:
[15:02:23]     [echo] using custom credentials
[15:02:23]
[15:02:23]prepare-tags:
[15:02:23]
[15:02:23]addTags:
[15:02:23]     [http] HTTP Request
[15:02:23]     [http] ********************
[15:02:23]     [http] URL:		https://buildserver/httpAuth/app/rest/8.0/builds/id:1302210/tags
[15:02:23]     [http] Method:		POST
[15:02:23]     [http] 	Content-Type: application/xml
[15:02:23]     [http] Entity:		yes
[15:02:23]     [http] ------ BEGIN ENTITY ------
[15:02:23]     [http] <tags><tag>ObfuscatedTag1</tag><tag>ObfuscatedTag2</tag></tags>
[15:02:23]     [http] ------- END ENTITY -------
[15:02:23]     [http] 
[15:02:23]     [http] HTTP Response
[15:02:23]     [http] ********************
[15:02:23]     [http] Status:		401
[15:02:23]
[15:02:23]BUILD FAILED
[15:02:23]E:\Obfuscated\Obfuscated\build9079597898286185054.tmp:73: Expected Status [204] but got [401] for URI [https://buildserver/httpAuth/app/rest/8.0/builds/id:1302210/tags]

I can't figure out the cause. Credentials in the build steps do not change between builds. Is using "8.0" in the URL a problem? Should it use the most recent endpoint?

Is it possible that TeamCity cannot handle these REST calls if multiple calls happen at the same time?

Hello,
since the release of 2.3 of XUnit, the meta-runner does not work anymore. The XUnit team changed the location of the xunit.runner.console.

The xunit.runner.console NuGet package has been changed; notably, the path to the net452 executables has moved from tools\ to tools\net452. You will need to update any paths which previously hard-coded this path.

You guys will have to check the version of XUnit downloaded from the feed to determine if the console application is in tools or tools\net452.

Also, I noticed that after the changes that @asif Ahmed on supporting multiple paths, the script gives me the the following errors:

##teamcity[buildStatus text='Method invocation failed because [System.Object[]
[13:14:37]
[Step 1/1] Value should end with "'"
Valid property list format is (name( )*=( )*'escaped_value'( )*)* where escape symbol is "|"
[13:14:37][Step 1/1] ] does not contain a method named 'Trim'.' status='FAILURE']
[13:14:37][Step 1/1] ##teamcity[message text='Method invocation failed because [System.Object[]

I do not have any problems using commit d310194.

Thank you.

The Gerrit Metarunner always votes +1, verified, even if the build fails?

Expected:
If the Build is failing, -1 is voted!

Hi,

All in title : i can't get the PHPUnit metarunner work with code coverage :

[12:22:11]Step 2/2: Run unit tests & coverage (PHP - PHPUnit) (3s)
[12:22:11][Step 2/2] Step 1/1: Ant (3s)
[12:22:13][Step 1/1] collectCoverage
[12:22:13][collectCoverage] zip
[12:22:13][zip] /home/teamcity/TeamCity/buildAgent/work/88c8db464164b0a9/tccoverage does not exist.
[12:22:14][Step 1/1] Step Ant failed

I'm using the dev-master branch

Hi there,
I was facing an issue using the MRPP_PHP_PHPUnit because the project I was trying to test didn't have a configuration file but a test folder with php scripts.

The test where failing because of missing xml file and to change the runner this way:
<arg line="-c &quot;${phpunit.configuration}&quot; --log-junit ${phpunit.junit.log} ${phpunit.coverage.param} ${phpunit.params}"/>
to
<arg line="&quot;${phpunit.configuration}&quot; --log-junit ${phpunit.junit.log} ${phpunit.coverage.param} ${phpunit.params}"/>

Is there any other way I can acchieve the same result without modifying the runner?

Thanks
Best regards
Andrea

When running coverage on tests that use SQL Server, this error appears:

[JetBrains dotCover] Coverage session finished with errors: Unable to profile the application because the host protection is applied. Probably it relates with Microsoft SQL; Unable to profile the application because the host protection is applied. Probably it relates with Microsoft SQL.

This issue and a workaround (add /ProcessFilters=-:sqlservr.exe to dotCover.exe arguments) is describe here: https://teamcity-support.jetbrains.com/hc/en-us/community/posts/205410670-Can-t-generate-the-code-coverage-report

Hello,

How can I solve the following issue in a build?

[11:35:07][Step 1/1] getPhpUnit (1s)
[11:35:07][getPhpUnit] get (1s)
[11:35:07][get] Getting: http://phar.phpunit.de/phpunit.phar
[11:35:07][get] To: /opt/TeamCity/buildAgent/temp/buildTmp/phpunit.phar
[11:35:08][get] http://phar.phpunit.de/phpunit.phar permanently moved to    https://phar.phpunit.de/phpunit.phar
[11:35:09][get] Error getting http://phar.phpunit.de/phpunit.phar to /opt/TeamCity/buildAgent/temp/buildTmp/phpunit.phar
[11:35:09][get] javax.net.ssl.SSLHandshakeException: sun.security.validator.ValidatorException: PKIX path building failed: sun.security.provider.certpath.SunCertPathBuilderException: unable to find valid certification path to requested target

[15:02:21]Step 6/6: Deploy War (Deploy War) (3s)
[15:02:21][Step 6/6] Step 1/1: Ant (3s)
[15:02:24][Step 1/1] deploy
[15:02:24][deploy] cargo
[15:02:25][cargo] org.codehaus.cargo.container.ContainerException: Failed to create deployable with implementation class org.codehaus.cargo.container.tomcat.TomcatWAR for the parameters (container [id = [tomcat6x]], deployable type [war]).
[15:02:25][Step 1/1] Process exited with code 1
[15:02:25][Step 1/1] Ant output
[15:02:25][Ant output] at org.apache.tools.ant.Target.performTasks(Target.java:411)
[15:02:25][Ant output] at org.apache.tools.ant.Project.executeSortedTargets(Project.java:1399)
[15:02:25][Ant output] at org.apache.tools.ant.Project.executeTarget(Project.java:1368)
[15:02:25][Ant output] at org.apache.tools.ant.helper.DefaultExecutor.executeTargets(DefaultExecutor.java:41)
[15:02:25][Ant output] at org.apache.tools.ant.Project.executeTargets(Project.java:1251)
[15:02:25][Ant output] at org.apache.tools.ant.Main.runBuild(Main.java:809)
[15:02:25][Ant output] at org.apache.tools.ant.Main.startAnt(Main.java:217)
[15:02:25][Ant output] at org.apache.tools.ant.launch.Launcher.run(Launcher.java:280)
[15:02:25][Ant output] at org.apache.tools.ant.launch.Launcher.main(Launcher.java:109)
[15:02:25][Ant output] Caused by: java.lang.NoClassDefFoundError: org/jdom/Parent
[15:02:25][Ant output] at java.lang.ClassLoader.defineClass1(Native Method)
[15:02:25][Ant output] at java.lang.ClassLoader.defineClass(ClassLoader.java:792)
[15:02:25][Ant output] at org.apache.tools.ant.AntClassLoader.defineClassFromData(AntClassLoader.java:1124)
[15:02:25][Ant output] at org.apache.tools.ant.AntClassLoader.getClassFromStream(AntClassLoader.java:1295)
[15:02:25][Ant output] at org.apache.tools.ant.AntClassLoader.findClassInComponents(AntClassLoader.java:1351)
[15:02:25][Ant output] at org.apache.tools.ant.AntClassLoader.findClass(AntClassLoader.java:1311)
[15:02:25][Ant output] at org.apache.tools.ant.AntClassLoader.loadClass(AntClassLoader.java:1064)
[15:02:25][Ant output] at java.lang.ClassLoader.loadClass(ClassLoader.java:357)
[15:02:25][Ant output] at org.codehaus.cargo.module.webapp.tomcat.TomcatWarArchive.parseTomcatContextXml(TomcatWarArchive.java:92)
[15:02:25][Ant output] at org.codehaus.cargo.module.webapp.tomcat.TomcatWarArchive.(TomcatWarArchive.java:59)
[15:02:25][Ant output] at org.codehaus.cargo.container.tomcat.TomcatWAR.(TomcatWAR.java:50)
[15:02:25][Ant output] ... 29 more
[15:02:25][Ant output] Caused by: java.lang.ClassNotFoundException: org.jdom.Parent
[15:02:25][Ant output] at org.apache.tools.ant.AntClassLoader.findClassInComponents(AntClassLoader.java:1361)
[15:02:25][Ant output] at org.apache.tools.ant.AntClassLoader.findClass(AntClassLoader.java:1311)
[15:02:25][Ant output] at org.apache.tools.ant.AntClassLoader.loadClass(AntClassLoader.java:1064)
[15:02:25][Ant output] at java.lang.ClassLoader.loadClass(ClassLoader.java:357)
[15:02:25][Ant output] ... 40 more
[15:02:25][Ant output]
[15:02:25][Ant output] Total time: 3 seconds
[15:02:25][Step 1/1] Step Ant failed

I'm using the meta runner and everything is ok, but I'm not able to use new C# syntax because it seems that it is launching old version of msbuild.exe.

Location of Msbuild is c:\Windows\Microsoft.NET\Framework\v4.0.30319\msbuild.exe and in build machine there is VS2015 installed.

How can I configure Gitversion meta-runner to use the correct version of msbuild?

Gian Maria.

With the PHP meta runners installed on TeamCity 8.0.6, the ant script would fail even if phpunit had an exit code of 0:

[Step 1/1] runTests
[runTests] fail
[fail] Unit test error.
[Step 1/1] Process exited with code 1

We were able to fix this by changing the fail condition from

<fail message="Unit test error." if="phpunit.result.error" />

To

<fail message="Unit test error." if="${phpunit.result.error}" />

Is this a compatibility issue with the built-in Ant? Will updating the runner cause issues with older environments?

I was assuming that once I switched my checkout mode from Automatically on Server to Automatically on Agent, the repository + all git info would be pulled for GitVersion to work properly, but that was not the case for me.

The repo is pulled as expected (by TeamCity), but when I get to my GitVersion runner step, it says: Fetching from remote 'origin' using the the following refspecs: +refs/heads/:refs/remotes/origin/.

Then it fails: LibGit2Sharp.LibGit2SharpException: Request failed with status code: 401

Likely because the runner does not have the username.

When I go back to edit the steps properties, add the repo url, with the username and password it works as expected - but files are checked out twice, once by teamcity and then again by the meta runner. Is this to be expected? I thought once the files were checked out by teamcity and all the git info was available that GitVersion would have everything it needs to process. Am I missing something here?

What about bower support?
+1

We use the xUnit + dotCover as one of the build steps in a TeamCity configuration. The step started failing consistently around 7/17/2018 on some of the branches with the following message:
##teamcity[buildStatus text='Cannot find path 'C:\BuildAgent\work\a089c236b45ffff1\xunit.runner.console.2.3.1\tools\net452\xunit.console.exe C:\BuildAgent\work\a089c236b45ffff1\xunit.runner.console.2.4.0\tools\net452\xunit.console.exe.config' because it does not exist.' status='FAILURE']
We use git and 'Automatically on server' checkout mode for the version control.

Upon further investigation, we discovered the following:

• Meta runner installs the latest version of the xunit.runner.console nuget package into the working directory.
• Version 2.4.0 of the xunit.runner.console package came out around the same time our builds started failing.
• The working folders on all build agents had both xunit.runner.console.2.3.1 and xunit.runner.console.2.4.0 folders at the time of the failures.
• Meta runner code at MRPP_xunit_dotcover.xml#L78 is using Resolve-Path on xunit.runner.console.* which I believe returns an array of strings in case if both xunit.runner.console.2.3.1 and xunit.runner.console.2.4.0 folders are present.
• The meta runner also appends .config suffix to the path and then fails because the path 'C:\BuildAgent\work\a089c236b45ffff1\xunit.runner.console.2.3.1\tools\net452\xunit.console.exe C:\BuildAgent\work\a089c236b45ffff1\xunit.runner.console.2.4.0\tools\net452\xunit.console.exe.config' does not exist.
• Manually deleting xunit.runner.console.2.3.1 folder on a build agent fixes the problem.

Is this a bug in the meta runner code? Or is it something wrong with the configuration we use? It seems like meta runner code should account for having multiple version of the xunit.runner.console package in the working folder and only use the most recent one. Or even better, the version of the xunit.runner.console package should be a configurable parameter.

If xUnit Console NuGet package contains multiple xunit.console.exe files, current code will include them all in the command line causing the xUnit execution to fail. For example, the standard xunit.runner.console.2.4.1 package contains exe's for net542, net46, net461, net462, etc.

May want to add a parameter for specific framework version or at least check for multiple results and just grab the first one (that's what I ended up doing).

Original code:

## get xunit runner from nuget package feed
Invoke-Expression "$nugetExe install xunit.runner.console -source $xUnitNuget"
$xunit = Join-Path $workingDir "xunit.runner.console.*\tools\$xUnitExe" | Resolve-Path

if (-not $xunit) {
    # Try finding xunit under framework specific folder
    $xunit = Join-Path $workingDir "xunit.runner.console.*\tools\net4*\$xUnitExe" | Resolve-Path
}

Workaround with check for multiple results (also include fix for space in path mentioned in #135:

## get xunit runner from nuget package feed
Invoke-Expression "& '$nugetExe' install xunit.runner.console -source $xUnitNuget"
$xunit = Join-Path $workingDir "xunit.runner.console.*\tools\$xUnitExe" -Resolve
if ($xunit.Length -gt 0) {$xunit = $xunit[0]}

if (-not $xunit) {
    # Try finding xunit under framework specific folder
    $xunit = Join-Path $workingDir "xunit.runner.console.*\tools\net4*\$xUnitExe" -Resolve
    if ($xunit.Length -gt 0) {$xunit = $xunit[0]}
}

Hi,

We are using the extension; when running in our build, we are noticing the following error:

[Step 5/5] Step 1/1: Install Test runner, Create config, Run tests, Update coverage (Powershell) (4s)
[15:24:42][Step 1/1] PowerShell Executable: C:\Windows\SysWOW64\WindowsPowerShell\v1.0\powershell.exe
[15:24:42][Step 1/1] Working directory: C:\BuildAgent\work\7db49b1ba6aa4694
[15:24:42][Step 1/1] PowerShell arguments: [-NoProfile, -NonInteractive, -ExecutionPolicy, ByPass, -File, C:\BuildAgent\temp\buildTmp\powershell1783826750060090740.ps1]
[15:24:45][Step 1/1] Feeds used:
[15:24:45][Step 1/1] http://OUR_LOCAL_NEXUS_NUGET_FEED
[15:24:45][Step 1/1]
[15:24:45][Step 1/1] Unable to load the service index for source http://OUR_LOCAL_NEXUS_NUGET_FEED .
[15:24:45][Step 1/1] Key not valid for use in specified state.
[15:24:46][Step 1/1] No test assemblies found
[15:24:46][Step 1/1] Process exited with code 0

NuGet: 3.4.3.
TeamCity: 9.1.6
Compiler: Visual Studio 2015-SP2

Thanks pointers on how to fix this would greatly be appreciated.

-Long

When "Update GitVersion" is checked in the TeamCity build step configuration, I am seeing two warning in the build log that seem to be coming from Chocolatey.

One warning says:

[14:23:19][Step 1/1] DEPRECATION NOTICE - choco update is deprecated and will be removed or
[14:23:19][Step 1/1] replaced in version 1.0.0 with something that performs the functions
[14:23:19][Step 1/1] of updating package indexes. Please use choco upgrade instead.

Another says:

[14:23:42][Step 1/1] WARNING: Write-ChocolateySuccess is deprecated. If you are the maintainer,
[14:23:42][Step 1/1] please remove it from your package file.

I have Chocolatey v0.9.9.12 (currently the latest) installed.

Here are these messages in their context:

[14:23:15][Step 1/1] Checking for updated version of GitVersion
[14:23:19][Step 1/1]
[14:23:19][Step 1/1] DEPRECATION NOTICE - choco update is deprecated and will be removed or
[14:23:19][Step 1/1] replaced in version 1.0.0 with something that performs the functions
[14:23:19][Step 1/1] of updating package indexes. Please use choco upgrade instead.
[14:23:19][Step 1/1] Upgrading the following packages:
[14:23:19][Step 1/1] gitversion.portable
[14:23:19][Step 1/1] By upgrading you accept licenses for the packages.
[14:23:23][Step 1/1] You have GitVersion.Portable v3.4.1 installed. Version 3.5.4 is available based on your source(s).
[14:23:31][Step 1/1]
[14:23:31][Step 1/1] GitVersion.Portable v3.5.4
[14:23:42][Step 1/1] Added C:\ProgramData\chocolatey\bin\gfv.exe shim pointed to '..\lib\gitversion.portable\tools\gitversion.exe'.
[14:23:42][Step 1/1] WARNING: Write-ChocolateySuccess is deprecated. If you are the maintainer,
[14:23:42][Step 1/1] please remove it from your package file.
[14:23:44][Step 1/1] ShimGen has successfully created a shim for GitVersion.exe
[14:23:44][Step 1/1] The upgrade of gitversion.portable was successful.
[14:23:44][Step 1/1]
[14:23:44][Step 1/1] Chocolatey upgraded 1/1 package(s). 0 package(s) failed.
[14:23:44][Step 1/1] See the log for details (C:\ProgramData\chocolatey\logs\chocolatey.log).

I was trying to make a build chain B->A.
However I have a custom build in B. It turns out that A starts when B's custom build is still running.
How can fix that?

Thank you.

Should we introduce a new parameter to specify the XUnit version to be downloaded internally, by the plugin on TeamCity Server. The main idea is not to use the latest version of XUnit as I have observed multiple issues across XUnit and .NetFramework. Restricting to a specific version will make sure that no issues come up.

So, NuGet seems to provide an option to download the specific version of Console Runner.
<NugetExe> install xunit.runner.console -source <XUnitSource> -version 2.1.0

Perhaps this new parameter could be an OPTIONAL parameter, which if empty would imply that the latest version should be fetched from NuGet.

So,
If Parameter value exists
<NugetExe> install xunit.runner.console -source <XUnitSource> -version <Version>

If Parameter value is empty,
<NugetExe> install xunit.runner.console -source <XUnitSource>

Please suggest if this would be a good feature to be introduced within the plugin ?

[Custom Build] How to add property queueAtTop="true" to the build?

Thank you

TeamCity Professional 2018.1 (build 58245)
The error in "Step 1/1: Install Test runner, Create config, Run tests, Update coverage (PowerShell)":
[Step 1/1] Incorrect 'TargetExecutable' value 'C:\TeamCity\buildAgent\work\6fc5b5e66f19d062\xunit.runner.console.2.4.0\tools\net452\xunit.console.exe C:\TeamCity\buildAgent\work\6fc5b5e66f19d062\xunit.runner.console.2.4.0\tools\net46\xunit.console.exe C:\TeamCity\buildAgent\work\6fc5b5e66f19d062\xunit.runner.console.2.4.0\tools\net461\xunit.console.exe C:\TeamCity\buildAgent\work\6fc5b5e66f19d062\xunit.runner.console.2.4.0\tools\net462\xunit.console.exe C:\TeamCity\buildAgent\work\6fc5b5e66f19d062\xunit.runner.console.2.4.0\tools\net47\xunit.console.exe C:\TeamCity\buildAgent\work\6fc5b5e66f19d062\xunit.runner.console.2.4.0\tools\net471\xunit.console.exe C:\TeamCity\buildAgent\work\6fc5b5e66f19d062\xunit.runner.console.2.4.0\tools\net472\xunit.console.exe'

The problem is in the following code:

$xunit = Join-Path $workingDir "xunit.runner.console.*\tools\$xUnitExe" | Resolve-Path
    
if (-not $xunit) {
    # Try finding xunit under framework specific folder
     $xunit = Join-Path $workingDir "xunit.runner.console.*\tools\net4*\$xUnitExe" | Resolve-Path
}

When $xUnitExe is "xunit.console.exe" (as specified in readme.md) then Join-Path command will return multiple results for the "xunit.runner.console.\tools\net4\$xUnitExe" path.

As a fast solution can be specified the folder name for "Xunit Runner Executable" param:
"net461\xunit.console.exe".

would be cool if it supported artifacts rules