[Cucumber, Webrat] Unable to run 2 features in a row


#1

Hi,

I have written 2 features that each have 1 scenario. When I execute each
feature separately with “rake features FEATURE=features/…” they each
pass, but when I do “rake features”, the first feature passes, and the
second one fails.

In my Given steps, I populate the DB, and some Given steps are called by
both features. I can’t call Given in only one feature and not the other,
because in such case, individually running each feature would fail. So I
have added a check, if the entry already exists, then I shall not create
a duplicate. But it still doesn’t work.

How to get around this headache? Anyone already bumped into that?


#2

Fernando P. wrote:

Hi,

I have written 2 features that each have 1 scenario. When I execute each
feature separately with “rake features FEATURE=features/…” they each
pass, but when I do “rake features”, the first feature passes, and the
second one fails.

In my Given steps, I populate the DB, and some Given steps are called by
both features. I can’t call Given in only one feature and not the other,
because in such case, individually running each feature would fail. So I
have added a check, if the entry already exists, then I shall not create
a duplicate. But it still doesn’t work.

How to get around this headache? Anyone already bumped into that?

Okay I get it now. That was a nasty one:

Between each Scenario, the DB gets cleared, however when adding entries
to the DB, the object’s id gets incremented (i.e: it won’t reset to 1).
So you cannot do something like the following in your step files:
product.author_id = 1

The way to fix this problem is to instantiate the object as an instance
var:
@author = Author.create!

And then in the step definition you get its id using:
product.author_id = @author.id

It took me a few hours to figure out, and a lot of trial and error.

As I don’t see any other way around, I guess it is a good idea to update
the following documentation page accordingly to explain in which case an
instance var is compulsory:
http://wiki.github.com/aslakhellesoy/cucumber/step-organisation


#3

On 18 Jan 2009, at 00:07, Fernando P. wrote:

How to get around this headache? Anyone already bumped into that?

Okay I get it now. That was a nasty one:

Between each Scenario, the DB gets cleared, however when adding
entries
to the DB, the object’s id gets incremented (i.e: it won’t reset to
1).
So you cannot do something like the following in your step files:
product.author_id = 1

Yes, hard-coding database identifiers anywhere is going to bite you
somewhere painful every single time.

As I don’t see any other way around, I guess it is a good idea to
update
the following documentation page accordingly to explain in which
case an
instance var is compulsory:
http://wiki.github.com/aslakhellesoy/cucumber/step-organisation

Not necessarily compulsory. You can also do this:

Given “The Book was written by the Author” do
Product.count.should == 1
the_product = Product.first
Author.count.should == 1
the_author = Author.first
product.author_id = author.id
end

We have a convention on my team that as long as your step refers to to
the Widget, then

  • there should only be one Widget in the database at that moment
  • you can therefore assume that the feature is referring to that
    one and only Widget

Matt W.
http://blog.mattwynne.net
http://www.songkick.com