Cucumber::Ast::Table.diff! is broken when using no headers

[troessner]

I have the following lines in one of my scenarios:

And I should see locations table                                             
  | Name    |      |      |
  | Berlin  | Show | Edit |
  | Hamburg | Show | Edit |

The step definition for this is (debugger added, see later):

Then(/^I should see (.+) table$/) do |table_id, expected_table|
  html_table = table_at("##{table_id.parameterize.tableize}").to_a
  html_table.map! { |r| r.map! { |c| c.gsub(/<.+?>/, '').gsub(/[\n\t\r]/, '') } }
debugger
  expected_table.diff!(html_table)
end

Now, my table does exactly look like specified above (more no this later), but the step still fails:

Tables were not identical (Cucumber::Ast::Table::Different)
./features/step_definitions/common_steps.rb:18:in `/^I should see (.+) table$/'
features/manage_locations.feature:15:in `And I should see locations table'

The problem here is that some of the columns have an empty header and that's where the diff! method falls short.
Using the debugger I can see the following:

<(rdb:1) p expected_table.raw
[["Name", "", ""], ["Berlin", "Show", "Edit"], ["Hamburg", "Show", "Edit"]]
(rdb:1) p html_table
[["Name", "", ""], ["Berlin", "Show", "Edit"], ["Hamburg", "Show", "Edit"]]
(rdb:1) p expected_table.raw == html_table
true

So, as you can see, the "raw" representation of the expected_table is the very same as of the html_table but the step still fails.

I think there are two things wrong here:

1.) IMHO it shouldn't fail since the tables are identical regarding everything I can see.
2.) At least the error message should tell me what exactly is not identical. From the output above I have no chance to fix this.

I looked into the implementation of the diff!-method, but unfortunately this isn't really a beauty (no offense), so it's definitely no 5 minute job to improve this. (To be honest, I think the whole "diff!" method needs a huge refactoring - it's way to long, nothing is really abstract or loosely coupled and it's badly readable.)

Possible cause:

I think the problem is that the method calls hashes internally, which uses the table headers as keys, if more than one table header is not set -> boom, there goes one column.

A colleague of mine created a sample project to illustrate this: https://github.com/exviva/cucumber_table_diff

What do you think?
Am I missing something here?

[ramgole]

This also happens if 2 headers are the same like

  | Name    | X    | X    |
  | Berlin  | Show | Edit |
  | Hamburg | Show | Edit |

[ramgole]

I avoided this problem by calling transpose on both tables before taking diff!

[os97673]

@robertodecurnex are you going to work on this issue?

[robertodecurnex]

Yea, I'll be digging over this.

There are conceptual issues in the middle too :S

I'll try to figure it out and maybe jump into an IRC discussion if I can not.

[mattwynne]

Thanks @robertodecurnex!

[mattwynne]

@robertodecurnex how are you getting on? Need any help?

[mattwynne]

Closing due to inactivity. Please reopen if and when you have a fix.

[betelgeuse]

@mattwynne wouldn't it be better to keep bugs open so users are more likely to find them and pick them up to work on them?

[mattwynne]

@betelgeuse I don't like having bugs hanging around not being fixed. To me, this one is an edge-case that nobody on the core team is going to take care of. So I've closed it because if someone cares about it, they can re-open it and work on it.

But maybe you're right, and that's a confusing way for us to operate.

Would you like to work on this yourself?

[lock]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.