opensoul.org - Cucumber scenarios that depend on Sphinx Comments

Comment on 'Cucumber scenarios that depend on Sphinx' by Guillermo

2009-09-29T10:44:14Z

I don’t disable transactions, so i write a method to commit the status before index:

Given /^the search index was generated$/ do
  without_transactions do
    TS.controller.index
  end
end

without_transactions method just commit current transactions, yield, and create transactions again:

def without_transactions
  @__cucumber_ar_connection.open_transactions.times do
    @__cucumber_ar_connection.commit_db_transaction
  end
  yield
  @__cucumber_ar_connection.open_transactions.times do
    @__cucumber_ar_connection.begin_db_transaction
  end
end

Features that use “the search index was generated” step, should be tagged with @sphinx, so i can just do

Before('@sphinx') do
  TS.build
  FileUtils.mkdir_p TS.searchd_file_path
  TS.controller.start
end

After('@sphinx') do
  TS.controller.stop
  reset_database!
end

The reset_database! use without_transaction and do something similar to database_cleaner.

By this way, i can use transactions in all of my code. I just need to reset the database in features (or scenarios) tagged as @sphinx

Comment on 'Cucumber scenarios that depend on Sphinx' by bmsatierf

2009-09-08T17:02:54Z

Great!

Thank you very much.

Comment on 'Cucumber scenarios that depend on Sphinx' by schlick

2009-06-26T03:39:54Z

Alternatively, you can keep the database cleaning at the start of each scenario and simply perform a final clean at the end of all your tests using:

at_exit do
  DatabaseCleaner.clean
end

Comment on 'Cucumber scenarios that depend on Sphinx' by Erik Ostrom

2009-06-23T22:38:55Z

For my cucumber features I’ve been using fixtures, loaded from env.rb, and transactional fixtures to reset them after each scenario. Working from your post, I disabled transactional fixtures, and replaced them with DatabaseCleaner’s truncation. When I ran the scenarios, my fixtures were all gone!

I think this is because, well, I was deleting them before each scenario. From the DatabaseCleaner README, it seems the preferred approach is to clean after each scenario:

require 'database_cleaner'
DatabaseCleaner.strategy = :truncation
Before do
  DatabaseCleaner.start
end
After do
  DatabaseCleaner.clean
end

Clean-before will work with other fixture strategies, but I don’t think there’s a down side to clean-after.

Comment on 'Cucumber scenarios that depend on Sphinx' by schlick

2009-06-19T12:47:47Z

Yep, it really works now. Excellent work!

Comment on 'Cucumber scenarios that depend on Sphinx' by Tim Riley

2009-06-17T00:43:30Z

Thank you for this post! This has helped us finally provide coverage for the only part of our application that lacked cucumber features.

Comment on 'Cucumber scenarios that depend on Sphinx' by Brandon

2009-06-13T02:06:56Z

I’ve updated the post again. Now it really works without already having the test index generated. Let me know if you have any other issues.

Comment on 'Cucumber scenarios that depend on Sphinx' by schlick

2009-06-11T13:40:55Z

Yep, same as Matt. When no db/sphinx/test directory exists beforehand, running the cucumber tests will fail. I had to manually perform rake ts:index RAILS_ENV=test in order for this directory to be created and populated. I would’ve thought ThinkingSphinx::Configuration.instance.build would do it but it doesn’t appear to. Still, thanks for the great blog post – it was the solution I’ve been looking for!

Comment on 'Cucumber scenarios that depend on Sphinx' by Matt

2009-06-08T23:41:12Z

Thanks! I actually had to run an index manually before it would start or stop my test sphinx. Anyway, got it working. :)

Comment on 'Cucumber scenarios that depend on Sphinx' by Brandon

2009-06-05T13:20:19Z

Matt, Sorry, in the process of getting this to work, I must have run rake thinking_sphinx:configure RAILS_ENV=test, which would have generated the test config. I deleted mine locally and I get the same error. To fix it, add the following line in env.rb right before calling #start:

ThinkingSphinx::Configuration.instance.build

I’ve updated the post to include this.

Comment on 'Cucumber scenarios that depend on Sphinx' by Matt

2009-06-03T23:22:21Z

Thanks for the write up, this is something I’ve been trying to do for a while.

Unfortunately it says “Failed to start searchd daemon.” whenever I try to run my features. Did you have to change your sphinx config at all? Any other ideas?

Comment on 'Cucumber scenarios that depend on Sphinx' by Brandon

2009-06-02T17:44:14Z

Joe, thanks, I fixed the link. I’ll leave it up to the maintainer of Thinking Sphinx to decide if he wants to pull any of it.

Comment on 'Cucumber scenarios that depend on Sphinx' by Joe Van Dyk

2009-06-02T16:28:51Z

The database_cleaner link is broken.

Would it be possible to put that code into ThinkingSphinx itself? Perhaps it could run whenever it detects that cucumber is running?