Re: [geb-user] Ability to skip setup() for certain tests

[Alexander Kriegisch]

Hi Jeremy.
I agree with Marcin in everything he says. You really should refactor the way he suggested. And as he also said the first time he answered your question, this is a Spock rather than a Geb question.
But for what it is worth, I was a little bit bored just now and wanted to find out if an annotation-driven extension could solve the problem. Actually to quickly hack

an annotation @SkipSetup which you can add to any feature method and
a SkipSetupExtension extends AbstractAnnotationDrivenExtension with a visitFeatureAnnotation() method which gets triggered each time an annotated feature method is intercepted

is a matter of two minutes.
The complication here is the fact that actually nothing specific should happen in the annotated method itself but in the feature's setup() method which gets executed at another time (later, more precisely). So we have to find a way to communicate to the setup() method that we want to skip it. The way I did it is a bit contrived, but easy enough to implement:

The first time our annotation-driven extension gets triggered (i.e. the first time Spock finds a feature method annotated by @SkipSetup), we create a special method interceptor which only intercepts setup() methods.
We add that interceptor instance to the SpecInfo and also save a reference to it in the SkipSetupExtension in order to avoid creating a new interceptor each time we meet a @SkipSetup annotation. We only want to do that once.
The interceptor itself gets a List skippedFeatures property which we update with each newly found feature method name carrying the marker annotation.
During runtime the interceptor's interceptSetupMethod() method checks if it finds the current feature method name in its skippedFeatures list. Only if it does not, it proceeds to the setup method, otherwise it skips by just doing nothing.

Here is the source code for

the annotation,
the extension,
the interceptor and
two almost identical test classes (because we want to check if really one extension instance is created per specification so as not to get into trouble with our internal references) in order to check if it works as expected:


package de.scrum_master.testingimport org.spockframework.runtime.extension.ExtensionAnnotationimport java.lang.annotation.Retentionimport java.lang.annotation.Targetimport static java.lang.annotation.ElementType.METHODimport static java.lang.annotation.RetentionPolicy.RUNTIME@Retention(RUNTIME)@Target(METHOD)@ExtensionAnnotation(SkipSetupExtension)@interface SkipSetup {}

package de.scrum_master.testingimport org.spockframework.runtime.extension.AbstractAnnotationDrivenExtensionimport org.spockframework.runtime.model.FeatureInfoclass SkipSetupExtension extends AbstractAnnotationDrivenExtension {  SkipSetupMethodInterceptor interceptor  @Override  void visitFeatureAnnotation(SkipSetup annotation, FeatureInfo feature) {if (!interceptor) {  interceptor = new SkipSetupMethodInterceptor()  feature.spec.addSetupInterceptor interceptor}interceptor.skippedFeatures << feature.name  }}

package de.scrum_master.testing
import org.spockframework.runtime.extension.AbstractMethodInterceptorimport org.spockframework.runtime.extension.IMethodInvocationclass SkipSetupMethodInterceptor extends AbstractMethodInterceptor {  List skippedFeatures = new LinkedList<>()  @Override  void interceptSetupMethod(IMethodInvocation invocation) throws Throwable {if (!skippedFeatures.contains(invocation.feature.name))  invocation.proceed()  }}

package de.scrum_master.testingimport spock.lang.Specificationclass SkipSetupTestA extends Specification {  def setup() {println "SkipSetupTestA -> setup"  }  def feature1() {println "SkipSetupTestA -> feature1"expect: true  }  @SkipSetup  def feature2() {println "SkipSetupTestA -> feature2"expect: true  }  def feature3() {println "SkipSetupTestA -> feature3"expect: true  }  @SkipSetup  def feature4() {println "SkipSetupTestA -> feature4"expect: true  }}

package de.scrum_master.testingimport spock.lang.Specificationclass SkipSetupTestB extends Specification {  def setup() {println "SkipSetupTestB -> setup"  }  def feature1() {println "SkipSetupTestB -> feature1"expect: true  }  @SkipSetup  def feature2() {println "SkipSetupTestB -> feature2"expect: true  }  def feature3() {println "SkipSetupTestB -> feature3"expect: true  }  @SkipSetup  def feature4() {println "SkipSetupTestB -> feature4"expect: true  }}

I am also attaching a ZIP file with the source code for your convenience. 
Let us complete the picture with the console log output when running both specifications together:

SkipSetupTestA -> setupSkipSetupTestA -> feature1SkipSetupTestA -> feature2SkipSetupTestA -> setupSkipSetupTestA -> feature3SkipSetupTestA -> feature4SkipSetupTestB -> setupSkipSetupTestB -> feature1SkipSetupTestB -> feature2SkipSetupTestB -> setupSkipSetupTestB -> feature3SkipSetupTestB -> feature4
I think this is 

Re: [geb-user] Ability to skip setup() for certain tests

[jc]

Wow, wonder why i don't remember that.  My mistake!

-- 
You received this message because you are subscribed to the Google Groups "Geb 
User Mailing List" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to geb-user+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/geb-user/378a1f3f-3328-44bb-9830-a91910f2da8e%40googlegroups.com.

Re: [geb-user] Ability to skip setup() for certain tests

[Marcin Erdmann]

Jeremy,

I thought that you asked this question once before and after searching
through the past messages to the list from you it turned out that I was
right. I will just link to my reply to your earlier email on the topic:
https://groups.google.com/d/msg/geb-user/7vIvuB64CFw/CvcUVnxmFQAJ.
Personally, I'd consider the need to skip the setup method on one or two
tests a smell as well as an indicator that these tests do not belong in the
spec in question and should probably be moved out to a separate spec.

Marcin

On Fri, May 1, 2020 at 6:50 PM jc  wrote:

> I think a good enhancement would be the ability to skip the setup() for
> particular tests.  For example I have 20 tests but 1 or 2 of them are
> slightly different and don't need the setup().  Perhaps a flag of
> setup: false
>
> on the method or something like that.  Just throwing the idea out there as
> I have had several time this has come up for me.  We can easily get around
> it by just creating a method of what we want and putting it in all the
> tests except the ones that don't need it but I thought maybe others have
> come across this.  Just want to throw an enhancement idea out there.
>
> --
> You received this message because you are subscribed to the Google Groups
> "Geb User Mailing List" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to geb-user+unsubscr...@googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/geb-user/306afc9f-0883-4a40-b16c-e786dbfde9a4%40googlegroups.com
> 
> .
>

-- 
You received this message because you are subscribed to the Google Groups "Geb 
User Mailing List" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to geb-user+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/geb-user/CA%2B52dQQt_i-1EESWL9fjO3ZMB5FwFXCaeFKb46nn4eg_R2oG%2Bw%40mail.gmail.com.