I am currently working on QTP for some GE projects and as per the req we are moving from qtp to Selenium RC
As i am a new baby in selenium kindly tell me each and every step staring from download selenium files from path till the complete script running part using RUBY client(As we decided to use ruby client)

Kindly give me the examples like where to keep the selenium server files ruby files and our script file and when and how to run the server and when and how to run the ruby scripts

We are trying this from last 10 days and found no success
Your reply will be appreciated

Regards
Test

** Invoke selenium:test_with_server_started (first_time)
** Execute selenium:test_with_server_started
rake aborted!
Since Selenium spawns an internal app server, we need ActiveRecord to be multi-threaded. Please set ‘ActiveRecord::Base.allow_concurrency = true’ in your environment file (e.g. test.rb).

Sure enough, I got past the error by adding the suggested line to config/environments/test.rb

The next problem that I ran into was that the selenium server was looking for a binary called “firefox-bin” in my path. I’m running firefox 3 on linux, and it’s binary is just called “firefox”. I fixed the problem by making a symbolic link: /usr/local/bin/firefox-bin -> /usr/lib/firefox-3.0.1/firefox

After that my test ran fine.

Has anybody found a good way to configure selenium-rc_fu to use specified firefox profiles? I would like to have some tests that check how my app behaves in a browser with javascript enabled, and other tests that verify that the app gracefully degrades without javascript support (e.g. use a firefox profile with javascript disabled.)

>…I was wondering if there was any way around having to delete
>and reload the fixtures between each test…

I don’t know of a way. The test scripts have full access to the rails data and simultaneously drive selenium to perform operations on a browser which is in turn talking to a web server. If things were transactional the test scripts would not be able to see the state of the data. The “Full-stack webapp testing with Selenium and Rails” talk mentioned at the beginning of the article probably covers this better.

> …you say that transactional fixtures did not cause a problem for you.

You are correct that transactional fixtures would break seleniumrc_fu’s operations. Transactional fixtures are turned off in the seleniumrc_fu test base class. My test suite now has a mix of tests, most of which use transactional fixtures, but the selenium tests do not.

These tests are slow but really useful. At Pivotal Labs we use continuous integration, which partially takes the sting out of long running tests. You can make a change and run the tests that verify the change and check in. The CI box will run the full test suite, including the selenium tests on multiple browsers (firefox, IE).

Sorry for taking so long to reply, I got side-tracked by other issues.

That’s cleared up one confusion, which is why you were loading your fixtures into your development environment. That makes good sense to be able to see what your tests will see when creating the tests.

Looking back my main question is about the running of the tests after they’ve been written. Basically, I was wondering if there was any way around having to delete and reload the fixtures between each test. Clearly one possibility is using transactional fixtures as with the unit tests, though I was under the (possibly mistaken impression) that transactional fixtures as implemented unobtrusively by Rails wouldn’t work with Selenium RC-style tests. It’s a different process running the tests (rake selenium:test) to the one altering the database through the Rails app (the mongrel server), so any changes made by the mongrel process won’t be visible to the test task and the rollback by the test task won’t rollback any changes by the mongrel process.

However, you say that transactional fixtures did not cause a problem for you. So does that mean, they just worked with Selenium tests without any need for additional configuration or setup?

After rereading your comment and rereading the article, I think you may have been referring to the section where I suggest loading your fixture data into your dev environment. You asked about transactional fixtures and as I said in my last comment this did not cause a problem for me but it might for you.

I’m not sure the article made the point clear enough so I’m going to try again. Your selenium tests will probably need data, and in the process of creating your selenium tests initially it is helpful to drive the application manually so that you can see exactly what the selenium test will see. The best way to do this is 1) use fixture data for your tests and 2) load that fixture data into your dev environment while creating your tests so that you can run your app and see it just as your selenium tests will see it.

I believe that the normal, non-transactional fixture flow is 1) the fixture table is cleared and 2) the fixture data loaded, and this is done for each fixture specified for your tests. Transactional fixtures are an optimization to this process, since the fixture loading and the test are all done in a transaction, which is simply rolled back at the end of the test.

I hope this helps.