I have two patches I'm working on that I would like to submit for consideration to the Abbot codebase maintainers.
One of them allows for the use of either Jakarta or GNU or JDK 1.4's regex implementations as a plugable architecture decision one can make at runtime, instead of compile time.
The other patch allows for a pretty-printed listing of all the components discovered when findComponentInHeirarchy() method throws an exception. I find this useful on retrofitting tests into existing systems (where TDD was not initially done, but has become a go-forward decision). Eventually, this should be pretty-printed in the format of a ASCII tree, but now it's just a simple list.
And how do you folks feel about having a assertComponentInHeirarchy() method that does the same thing as findComponentInHeirarchy() except does NOT throw an exception, but returns a boolean?
Lastly, when I run all the tests in the test dir, I get a lot of failures?!?! Since AllTests.java is gone, what could I be doing wrong? Most of the errors seem to stem from Costello stuff, where Scripts aren't being found.
Running without an AllTests.java suite causes my IDE (Eclipse) to churn a lot. Having an AllTests suite makes everything go faster.
Thanks for your thoughts and feelings on this.
-Nick Bauman
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
The ant "test" target is easier to maintain then a bunch of AllTests targets, which is why I got rid of abbot.UnitTestSuite. You should be able to invoke it directly from eclipse, although any notes on facilitating that integration would be welcome (I really think ant and eclipse should get together on their dependency trees and stop fighting one another).
As for test failures, what platform, VM version, etc are you on? I run the tests regularly on w32/xp, linux/x11 and OSX with 1.4.2 (although OSX has had a lot of VM-related failures since the 1.4.x series came out). linux has a few (< 10) failures that I haven't had the opportunity to track down yet.
w32 should have *no* failures.
If you do get a lot of failures, do they seem to have a common point of failure (e.g. focus errors, text input, drag/drop)?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I've put abbot.UnitTestSuite back in; I'd like to have that auto-generated from available tests with a *little* bit of ordering (it doesn't make sense to run the full suite if the robot or windowing tests fail). Any suggestions are welcome.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
BTW, abbot.UnitTestSuite can be run from Eclipse to avoid making eclipse scan for classes. It also has a property you can set to specify a required classname prefix for tests you want to include/filter.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I have two patches I'm working on that I would like to submit for consideration to the Abbot codebase maintainers.
One of them allows for the use of either Jakarta or GNU or JDK 1.4's regex implementations as a plugable architecture decision one can make at runtime, instead of compile time.
The other patch allows for a pretty-printed listing of all the components discovered when findComponentInHeirarchy() method throws an exception. I find this useful on retrofitting tests into existing systems (where TDD was not initially done, but has become a go-forward decision). Eventually, this should be pretty-printed in the format of a ASCII tree, but now it's just a simple list.
And how do you folks feel about having a assertComponentInHeirarchy() method that does the same thing as findComponentInHeirarchy() except does NOT throw an exception, but returns a boolean?
Lastly, when I run all the tests in the test dir, I get a lot of failures?!?! Since AllTests.java is gone, what could I be doing wrong? Most of the errors seem to stem from Costello stuff, where Scripts aren't being found.
Running without an AllTests.java suite causes my IDE (Eclipse) to churn a lot. Having an AllTests suite makes everything go faster.
Thanks for your thoughts and feelings on this.
-Nick Bauman
The ant "test" target is easier to maintain then a bunch of AllTests targets, which is why I got rid of abbot.UnitTestSuite. You should be able to invoke it directly from eclipse, although any notes on facilitating that integration would be welcome (I really think ant and eclipse should get together on their dependency trees and stop fighting one another).
As for test failures, what platform, VM version, etc are you on? I run the tests regularly on w32/xp, linux/x11 and OSX with 1.4.2 (although OSX has had a lot of VM-related failures since the 1.4.x series came out). linux has a few (< 10) failures that I haven't had the opportunity to track down yet.
w32 should have *no* failures.
If you do get a lot of failures, do they seem to have a common point of failure (e.g. focus errors, text input, drag/drop)?
I've put abbot.UnitTestSuite back in; I'd like to have that auto-generated from available tests with a *little* bit of ordering (it doesn't make sense to run the full suite if the robot or windowing tests fail). Any suggestions are welcome.
Any progress on those patches?
BTW, abbot.UnitTestSuite can be run from Eclipse to avoid making eclipse scan for classes. It also has a property you can set to specify a required classname prefix for tests you want to include/filter.
I still have them here, and I have been tracking forward development in your repository. I just have a few more touches to add.