I've got a custom ant script for building and other fun stuff. IntelliJ has very nice integration for handling Ant tasks. However, every time I run a task IntelliJ opens the script in my editor panel. It moves to the line of the task I'm running and then I close it; actually I now have it in its own panel, minimized to the max.
Is there anyway to disable this behavior? I just want the Ant script to run the task, pop up its own panel with some feedback (which it does) and then allow me to keep coding without having to close the script. Please tell me this is possible.
Man, I can't believe this only had 4 views, and two of those were me. Pretty weak.
Anyways, I've got a solution that might help someone else down the road. In the Messages window that pops up when the ant build starts there is an Auto Scroll to Source button. Make sure this button is unchecked.
Related
We are using TFS2015 VNext Run functional tests tasks to run our coded UI tests in VMs.
Our Code UI tests click controls on UI, and somehow the command shell of DTAExecutionHost.exe was in "Select" state which prevent the test run from finishing.
We had to log into the VM and manually un-select the DTAExecutionHost.exe to let the test run finish.
Any idea how to minimize the windows to prevent it being clicked by UI tests? Or if any workarounds to avoid the issue?
Thank you!
Angel
I managed to work around the issue by minimize all windows right after Test Agent being configured.
We have an app.exe that uses another 3rd party installation wizard app, InstallAnywhere, to install our application.
It is a text-based wizard - meaning a new shell is spawned (and the child process is executing some java.exe in order for its wizard steps to work) when
app.exe -i console
in the parent shell is executed.
Can pexpect be used in this case as a kind of question/answer interaction ?
We tried it with Hudson but it appears that Hudson can only monitor the parent process and not its child process. When Hudson job kicks off the python script, it starts but fails to begin at step 1 of the wizard - it hangs.
How can we overcome this ?
Thanks
If I'm reading this correctly, you want to interact with your executing installer (in this case, using pexpect) during a Hudson job, probably to do automated testing.
I'm thinking that expect may get you where you want to go if you call expect from a shell script (or batch file) that is, in turn, run from Hudson. The script would execute the tests, using your favorite flavor of expect, during which it either exits successfully or fails. Hudson would detect the failure, and you could then react accordingly. Your script, and any expect messages, would write to stdout, which Hudson would collect into your build log.
However, Flexera has a testing framework that allows you to test an installer via JUnit. Look at the bottom of this page. Or check out the JavaDocs.
This might give you more precision than an expect hack. You might even be able to write a Hudson plugin and do way with expect and shell scripts completely.
Update: after looking over the Javadocs, I found that the main class GUIAutomationFixture uses Java's Robot class. This means you probably won't be able to run this class on a headless server. You'll likely need either Windows, or a Linux/Unix box with a working X system.
I have problems with creating a job in jenkins. At the "Build" step I select Build "execute windows batch command", where I want to start a browser such as opera, google chrome portable or firefox portable, the command never terminates.View screenshot
For IE, firefox and chrome I do not encounter any problems
If someone has a solution. thank you in advance
What do you expect? You have launched an executable that shows a form, albeit in a session that can never have a console session attached.
Jenkins, by design, will sit there until the executable terminates - which generally happens when the form is closed by a user interacting with it. Since there is no GUI, this process can only ever terminate if someone kills it with Task Manager.
What are you actually trying to achieve? There's almost certainly a reasonable way to do it.
I have been trying to do a GUI test with DUnit which includes interacting with modal windows via a message loop timer system by #tomazy (see my earlier question for more details: (How) Can I use FutureWindows with standard file open dialogs? ).
My solution in the other question works fine when I run the tests manually, but when I run this within a continuous build system in TeamCity, it hangs when the dialog's OK message should be handled. The service that runs the tests has the permission "interact with desktop" set, and I have verified that the hanging happens ONLY if the OK is pressed (i.e. the CDN_FILEOK notification happens). I can close the dialog using WM_CLOSE, but this naturally does not cause the dialog to return an OK modal result, and thus is not usable.
If I can't get this to work, I'll probably have to modify the production code (target of test) to publish an event to give the filename and have the test hook into it and give it without a dialog, but I'd like to know what is causing this problem and preferably of course solve it without modifying the production code.
I'm using ant as my primary builder for a flex project in IntelliJ. Every time I build, IntelliJ runs the code analysis tool and finds a bunch of errors in my build file, which pops up the "Messages" window every time. Actually, the errors that it finds are complaining mostly about properties which are defined in a separate file, which it doesn't actually parse when running the analysis.
Edit: Additionally, I am making use of ant tasks which generate properties on the fly, which IntelliJ is unaware of.
Is there any way to turn off file inspection for ant build files during the build process?
That sounds weird. IDEA on the fly inspections which you get while editing the build.xml file should not pop-up when you actually build the project unless they are really errors and are reported by Ant to IDEA while the build file is executed.
IDEA can perform additional validation and report errors on Make for certain components (see Settings | Compiler | Validation), but it doesn't include Ant build files and is not triggered when you just run the Ant target.
If you don't want to see warnings from the Ant Messages Window, you need to toggle off the "Show All Messages" button.
If you don't want IDEA to analyze your build.xml file for errors, click the Hector icon in the status bar and disable the inspections per file by changing the highlighting level from the Inspections to Syntax.
Clarify the problem if it's not the case (knowing IDEA version would be also useful).
Does the .properties file exist, and is it in your CLASSPATH? IntelliJ is very smart about figuring that stuff out. When IntelliJ complains, I usually assume that I'm making a mistake. That posture usually gets me to a solution pretty quickly.
You can disable Ant plugin altogether. IDEA becomes quite slow on large Ant builds, so found this to be of help:
Crtl+Alt+S (for settings) -> Plugins -> Ant Support (uncheck)