I am developing a server with the latest Go GAE SDK. I am running tests after every change:
goapp test -test.v=true
I am using -cover to record coverage as described by goapp help testflag:
goapp test -cover -test.v=true -test.coverprofile=c.out
[..]
coverage: 53.8% of statements
ok _/var/lib/jenkins/jobs/loyalty/workspace 30.464s
This completes successfully and prints the percentage of lines covered by tests. However, attempting to visualize the results fails:
goapp tool cover -html=c.out
cover: can't find "app.go": cannot find package "_/home/ingo/git/loyalty/" in any of:
/home/ingo/Downloads/go_appengine_sdk_linux_amd64-1.9.10/go_appengine/goroot/src/pkg/_/home/ingo/git/loyalty (from $GOROOT)
/home/ingo/git/loyalty/src/_/home/ingo/git/loyalty (from $GOPATH)
Does Go's cover tool only work on non-GAE apps? Do I have to package my app differently in order to visualize the coverage results?
I unsuccessfully asked this on golang-nuts before.
There is an open issue related to it. As a temporary workaround, I am running sed in between collecting and visualizing the coverage results.
goapp test -cover -test.v=true -test.coverprofile=c.out
sed -i -e "s#.*/\(.*\.go\)#\./\\1#" c.out
goapp tool cover -html c.out -o coverage.html
Related
I'm tying to write tests for my google cloud app. I read the documents and it seems the only way to run the test locally is the running the command goapp test in the test package directory. But when I run the command I get the error go: GOPATH entry is relative; must be absolute path: "".
I'm pretty sure my GOPATH is set absolutely. Here are the results when I run the command go env | grep GOPATH:
GOPATH=":/home/mohammad/go:/home/mohammad/go/src/gitlab.com/gc-auth"
Also getting the same output when I run echo $GOPATH.
Any help is appreciated.
PS: I have ubuntu 18.04 and my go version is 1.10.4
results of gcloud version:
Google Cloud SDK 228.0.0
app-engine-go
app-engine-python 1.9.80
bq 2.0.39
cloud-datastore-emulator 2.0.2
core 2018.12.07
gsutil 4.34
GOPATH=":/home/mohammad/go:/home/mohammad/go/src/gitlab.com/gc-auth"
The GOPATH starts with an empty path (which is technically relative), followed by two absolute paths. The error message is correct.
Is there any way to run the compiler on an App Engine application written in Go without continue to serve the application with the development server and instead get an exit code?
Because I want to add a check in my automated tests in Travis that the application actually compiles.
To clarify: I have access to the App Engine SDK / Development Server in Travis, but I dont want to run goapp serve since it never exits.
Without actually implementing test, your solution looks pretty hacky. Why not use goapp build? Here's my .travis.yml:
language: go
go:
- 1.2.1
# Grab newest version and suck down
install:
- export FILE=go_appengine_sdk_linux_amd64-$(curl https://appengine.google.com/api/updatecheck | grep release | grep -o '[0-9\.]*').zip
- curl -O https://storage.googleapis.com/appengine-sdks/featured/$FILE
- unzip -q $FILE
# Run build and tests
script:
- ./go_appengine/goapp test ./tests; # If you are testing
- ./go_appengine/goapp build ./packagedir; # Wherever you keep your stuff
For reference on tests or just to see a project that builds
Edit:
It has been awhile, but I noticed recently that some of my builds randomly break. It is infuriating and I have occasionally hardcoded SDK values to overcome this. No more. Here's a very hacky implementation of grabbing the first featured (and thus hosted as /updatecheck fails to always return a hosted version) of the SDK desired:
export FILE=$(curl https://storage.googleapis.com/appengine-sdks/ | grep -o 'featured/go_appengine_sdk_linux_amd64-[^\<]*' | head -1)
For just the file:
export FILE=$(curl https://storage.googleapis.com/appengine-sdks/ | grep -oP '(?<=featured/)go_appengine_sdk_linux_amd64-[^\<]*' | head -1)
I solved this by adding an empty Unit test at the entry point of the application (main_test.go). This unit test will force the whole application to compile.
Then I execute all unit tests by putting goapp test ./... in the script section.
Are there any third party tools compatible with robot framework to run the tests in three different browsers.
That means first it should run whole suite using IE
After that it should run whole suite using FF
Then using Chrome and this should be done automatically.
Please suggest me.
If you use Selenium2Library, and define the browser in a variable, you can run the same tests against multiple browsers by changing the variable on the command line:
$ pybot -v BROWSER=chrome my_test_suite.txt
$ pybot -v BROWSER=ff my_test_suite.txt
...
https://robotframework.org/#libraries has a nice list of good libraries for RF and among them there is a Selenium2Library.
I am running dev_appserver.py and so it auto builds as I save go files (I need to run this and not goapp because of log_level I need).
When there is a successful build I would like the tests for the project (goapp test) to run automatically. How can I do this?
You could use something like https://github.com/nf/watch in a separate terminal window. It'll re-run tests in parallel to dev_appserver.py.
Install: go get github.com/nf/watch
Run from your app's directory: watch goapp test
With AppEngine, I use the entr app like so:
$ find ./*.go | entr goapp test
What is the best way to run silverlight unit tests automatically using team city?
I have found StatLight which we had working well when we used cc.net, and it says that it has support for teamcity. Does this just mean the test results output file is compatible with teamcity? Do I need to create a command line runner to run the tests? If so how do I get the test results into team city?
Thanks
TeamCity has an extensibility feature where you can output special commands to the console and the TeamCity agent will capture the commands and publish the results within TeamCity.
If you get StatLight running on your desktop - do a regular console-run. Then do another run by giving it the "--teamcity" parameter. Notice the difference in the output?
In TeamCity you can setup a Command Line Build Runner
Command executable: "<Path to statlight.exe>"
Command parameters: "-x=%system.teamcity.build.checkoutDir%\PathToXap\SilverlightClient.Tests.xap --teamcity"
Hope this helps.
There's also a StatLight TeamCity plugin that adds a test runner.
You can try using Lighthouse Silverlight Unit Test Runner, it works with every Build Server including TeamCity and CCNet because it by default produces NUnit compatible xml results file:
http://lighthouse.codeplex.com/