03:32 <robholland> tomtt: hey :)

04:01 <robholland> aslakhellesoy: hey, seen my direct http call thread on the capybara ml? thoughts?

04:01 <aslakhellesoy> robholland: have seen it, but haven't given it much thought

04:02 <robholland> k

04:02 <aslakhellesoy> robholland: do your proposed changes only work with rack-test?

04:03 <robholland> yes, the other drivers don't support direct http calls

04:03 <robholland> (because they emulate drivers)

04:03 <robholland> er, browsers

04:03 <aslakhellesoy> so why not just rely on the rack-test API for HTTP?

04:03 <aslakhellesoy> i.e. let the user use that

04:03 <aslakhellesoy> user being the programmer

04:03 <robholland> because the capybara step definitions are useful when testing apis for us

04:04 <robholland> and I don't want to change the language to avoid the capybara ones

04:04 <robholland> and with the addition of those methods, I don't have to write any further code

04:04 <aslakhellesoy> looking at http://github.com/robholland/capybara/compare/master...headers

04:04 <robholland> it's all pragmastism rather than"design"

04:04 <robholland> Not the best patch tbh, should have made sure it was specifically the direct call methods

04:05 <robholland> The delegation stuff is just a tidy up obviously

04:05 <robholland> And the cache change is actually a separate (although required) change for the direct calls to work

04:05 <robholland> It actually has another benefit in that it prevents building the repsonse unless a step needs it

04:06 <robholland> Perhaps a @api patch is cleaner

04:06 <aslakhellesoy> The change is fairly small. Not a big deal that you clean up delegators in the same changeset.

04:06 <aslakhellesoy> I think it would make sense to add this to Capybara.

04:06 <robholland> A nod on the ml would be nice *puppy dog eyes*

04:07 <aslakhellesoy> As long as the documentation clearly states that this is doing things the user can't do.

04:07 <aslakhellesoy> robholland: will do

04:07 <robholland> ta

04:09 <aslakhellesoy> robholland: but if you're testing an API that, say, returns JSON. How would Capybara handle the result?

04:10 <aslakhellesoy> You can get the body allright, but wouldn't it also try to fit it into a Nokogiri DOM?

04:10 <aslakhellesoy> (and blow up)?

04:10 <robholland> Seemingly not, in that those stories pass

04:15 <aslakhellesoy> robholland: I would add a test that tries to click a link after getting a JSON resource.

04:15 <aslakhellesoy> -And verify that the error you get is something along the lines of "The last response was not HTML, but application/json."

04:16 <robholland> ok

04:17 <aslakhellesoy> Only #body (and maybe #headers) should work after a non-html response.

04:17 <robholland> That is a different issue though right?

04:17 <robholland> I mean, any link could result in a non-html response

04:27 <aslakhellesoy> robholland: true. so capybara should probably be behaving that way regardless of whether your change is merged in or not.

04:28 <robholland> indeed

05:55 <mfojtik> hi, just quick question, is there any way howto disable automatic loading of support/*/*.rb files ?

06:08 <aslakhellesoy> mfojtik: explicitly --require what you want loaded

06:08 <aslakhellesoy> --help for more

07:42 <flov> Anybody has some experience using capybara ?

07:44 <robholland> Aye

07:45 <flov> i have some trouble using the selectors. It just doesnt find that input field

07:45 <flov> i always get a Capybara::ElementNotFound exception Aaaargh

07:46 <robholland> What does the step look like?

07:52 <flov> robholland: http://pastie.org/873751

07:54 <flov> you can see the signup page at www.giveawaythings.org/signup

07:56 <robholland> flov: the label for the field is "Confirm it"

07:56 <robholland> Not Password Confirmation

07:56 <flov> well yes, ok, but if i set it to password_confirmation it doesnt give me a passing step neither

07:58 <flov> also, capybara just goes over the labels to select the input field? the error message says cannot fill in, no text field, text area or password field with *id, name, or label* 'Password Confirmation'

07:58 <robholland> that's because the field name is person[password_confirmation]

07:58 <robholland> And the id is person_password_confirmation

07:58 <robholland> Read the source :)

08:02 <flov> sorry, but which source? The Readme on github doesnt really explain how the fill_in method works

08:02 <robholland> The source code to the page

08:03 <robholland> You can use any of: 'person[password_confirmation]' or 'person_password_confirmation' or 'Confirm It'

08:03 <robholland> 'Confirm it' sorry (lower case i)

08:04 <robholland> It does what the error messages says, it tries what you've given it against the id, name and label for the field

08:04 <robholland> None of which are "Password Confirmation" or "password_confirmation"

08:08 <flov> robholland: well, i tried confirm it before but it just doesnt work

08:09 <flov> cannot fill in, no text field, text area or password field with id, name, or label 'Confirm It' found (Capybara::ElementNotFound)

08:09 <robholland> Lower case i

08:09 <robholland> Confirm it

08:10 <flov> cannot fill in, no text field, text area or password field with id, name, or label 'Confirm it' found (Capybara::ElementNotFound)

08:10 <flov> neither

08:10 <flov> but person[password_confirmation] worked

08:12 <robholland> You're label tag is incorrect

08:12 <robholland> for="person_confirm_it"

08:13 <robholland> If you fixed that, 'Confirm it' would work

08:13 <robholland> The label tag currently points to a nonexistent field name

08:17 <flov> ah, so the for attribute of the label tag points to the id tag of the input field?

08:17 <robholland> Yes. That is it's purpose

08:18 <robholland> Otherwise there would be no way to link a label with an input

08:18 <flov> well, you could take the next input attribute from the label tag on

08:19 <flov> i would be glad if that would be explained somewhere in the Readme of capybara. Did you actually read the code of capybara in order to understand whats happening?

08:19 <robholland> Hmm? The fact a label's for attribute must point to a valid input is nothing to do with capybara

08:19 <robholland> It's a requirement of HTML

08:21 <flov> ok, i didnt know what the label tag was actually about

08:23 <flov> robholland: thank you very much you saved my day :) i appreciate your help

08:24 <robholland> np

09:26 <kurkale6ka> Hi, I keep having this error: The WWW constant is deprecated, please switch to the new top-level Mechanize

11:14 <GitHub63> gherkin: 07eof-event 03ghnatiuk * 12b24c5 (15 files in 8 dirs): change message to eof and add to java

11:14 <GitHub63> gherkin: 07eof-event 03ghnatiuk * 296fd24 (4 files in 4 dirs): parser should allow empty feature files

11:14 <GitHub63> gherkin: 07eof-event 03aslakhellesoy * 2be4701 (26 files in 6 dirs): Add Eof() event to C#

11:14 <GitHub63> gherkin: 07eof-event commits f8ae002...2be4701 - http://bit.ly/djTOfl

11:15 <GitHub120> cucumber: 07treetop-behaves_like_gherkin 03aslakhellesoy * 5f4e051 (20 files in 5 dirs): Treetop parser pretty much conforms to the Gherkin API. Getting closer to drop in the Gherkin parser.

11:15 <GitHub120> cucumber: 07treetop-behaves_like_gherkin 03aslakhellesoy * c50324e (8 files in 2 dirs): Remove the old build methods in the Treetop parser.

11:15 <GitHub120> cucumber: 07treetop-behaves_like_gherkin commits f4c7640...c50324e - http://bit.ly/9Wr2gQ

11:35 <ghnatiuk> aslakhellesoy: hey

11:35 <ghnatiuk> not sure if you noticed http://bit.ly/cF94Wa

11:35 <ghnatiuk> or http://bit.ly/bL4s2Z

11:36 <ghnatiuk> but I removed the 'eof' from the 'Expected one of ...' parsing error message for ruby and java

11:36 <ghnatiuk> seemed like a strange thing to claim is expected

11:37 <aslakhellesoy> ghnatiuk: I might have missed something there.

11:37 <aslakhellesoy> so we don't need to expect the eof you mean?

11:37 <ghnatiuk> i think (based on the spec changes) that C# will still claim it's looking for eof

11:38 <ghnatiuk> well, up for debate :) "Expecting one of scenario, scenario_outline, eof, examples..."

11:38 <ghnatiuk> vs. leaving out the eof from that message

11:38 <GitHub77> cucumber: 07treetop-behaves_like_gherkin 03aslakhellesoy * 62eaca4 (4 files in 1 dirs): Use eof event - http://bit.ly/b9puS4

11:39 <aslakhellesoy> Oh I see. Yes, that's a minor detail we should fix in C#

11:40 <aslakhellesoy> I just pushed some code that makes Treetop behave (I think) exactly like Gherkin

11:40 <aslakhellesoy> About to see if I can drive Cucumber with Gherkin now.

11:40 <aslakhellesoy> Really excited

11:40 <mabes> that is awesome

11:41 <msassak> go go go go

11:41 <ghnatiuk> yay!!!

11:41 <aslakhellesoy> I felt really dirty when I refactored treetop and the ast

11:41 <msassak> preeeetty crufty in there

11:41 <aslakhellesoy> hell yeah

11:41 <aslakhellesoy> refs going left and right in all directions.

11:41 <aslakhellesoy> but it seems to work :-)

11:42 <aslakhellesoy> ahh shit

11:42 <msassak> what's wrong?

11:43 <aslakhellesoy> we have to move the filters into the builder too - they're stuck in treetop

11:43 <msassak> ugh

11:43 <aslakhellesoy> you know, only build ast based on tag and line filters

11:43 <aslakhellesoy> if present

11:44 <aslakhellesoy> well, shouldn't bee too much work

11:44 <aslakhellesoy> famous last words

11:44 <msassak> hah

11:44 <lucasbz> hey, can someone help me out on sth?

11:44 <msassak> lucasbz don't ask to ask :-)

11:44 <lucasbz> lol

11:46 <lucasbz> ok, using webrat's visit on a step when I'm rendering the corresponding view, I get an 'undefined method' error, it seems that this is happeing due to me using 'form_for :user' + text_field :name

11:46 <lucasbz> did I made myself any clear?

11:46 <lucasbz> :P

11:47 <aslakhellesoy> aaaaaaaarrrrrrrgghhh

11:47 <* aslakhellesoy> is just talking to himself

11:47 <ghnatiuk> uh oh

11:48 <aslakhellesoy> ghnatiuk: now we need to cache a lot more in the builder

11:48 <msassak> lucasbz can you gist the backtrace, feature and step definition?

11:48 <ghnatiuk> ugh. I can hardly wait for a new ast.

11:48 <ghnatiuk> that we can just apply filters to

11:48 <aslakhellesoy> yeh

11:49 <msassak> lucasbz err not the step def, but the view that's not being rendered

11:53 <lucasbz> msassak http://gist.github.com/335510

11:53 <lucasbz> :)

11:54 <msassak> have you run the migrations in the cucumber environment?

11:55 <lucasbz> hmm maybe not, dammit!

11:55 <lucasbz> thanks

11:55 <msassak> is that it?

11:56 <aslakhellesoy> so regarding filtering - cucumber has historically done it in 2 different ways

11:56 <lucasbz> yeah! unfortunately, or not...

11:57 <lucasbz> thank you again ;)

11:57 <aslakhellesoy> currently, if a filter is present (--tags, --name or a colon line number) the AST is only built up to match those

11:57 <msassak> lucasbz cool glad to be of help

11:58 <aslakhellesoy> previously (about a year ago) we'd always build the entire AST, but filter during traversal/execution

11:58 <aslakhellesoy> I don't remember why we switched

11:58 <aslakhellesoy> Probably performance related

11:58 <msassak> that would be my guess

11:58 <aslakhellesoy> It seems unnecessary to build nodes that never get executed

11:59 <aslakhellesoy> So I'm thinking we should go with the current approach

11:59 <aslakhellesoy> i.e. construct the builder with a filter, and ask the filter whether or not to create a node when we get events.

11:59 <aslakhellesoy> wdyt?

11:59 <msassak> i think that's best right now

12:00 <aslakhellesoy> ok, i'll go with that

12:00 <msassak> stick to how it is, rather than rewriting and integrating new things

12:01 <aslakhellesoy> msassak: you talkin to me?

12:01 <msassak> aslakhellesoy: yes, sorry :-)

12:01 <aslakhellesoy> hehe

12:01 <msassak> more rambling out loud, i suppose

12:07 <aslakhellesoy> msassak: nono, i need to hear it :-)

12:21 <aslakhellesoy> msassak_: ghnatiuk : look at this spec: http://github.com/aslakhellesoy/cucumber/blob/master/spec/cucumber/parser/feature_parser_spec.rb#L363

12:22 <aslakhellesoy> it parses this feature (specifying line 12): http://github.com/aslakhellesoy/cucumber/blob/master/examples/self_test/features/outline_sample.feature#L12

12:22 <aslakhellesoy> which means that row 10 and 11 should be omitted

12:22 <aslakhellesoy> and the entire Examples section of row 13 and below

12:23 <aslakhellesoy> That's a little tricky with a Gherkin based builder :-/

12:24 <aslakhellesoy> Oh, the scenario at line 3 should also be excluded

12:26 <aslakhellesoy> Maybe we can have 2 chained Gherkin builders?

12:26 <aslakhellesoy> A first one that stacks up all events and replays it to a 2nd one that builds the AST?

12:27 <aslakhellesoy> The 1st one could do the filtering before replaying the events maybe

12:27 <aslakhellesoy> wdyt?

12:28 <aslakhellesoy> this filtering/replaying listener could go into gherkin...

12:28 <ghnatiuk> yeah, I think passing the builder through a second filter is a good idea

12:28 <aslakhellesoy> filter_listener = Gherkin::FilterListener(tags=..., lines=..., names=..., listener)

12:28 <ghnatiuk> applying that to the treetop output

12:28 <ghnatiuk> in theory

12:29 <ghnatiuk> should leave it unchanged since treetop is filtering while parsing

12:29 <aslakhellesoy> ghnatiuk: it's trivial to remove the filtering from treetop

12:29 <ghnatiuk> oh, in gherkin itself?

12:29 <aslakhellesoy> -but we'd need a filtering listener in gherkin

12:30 <ghnatiuk> why not in cucumber?

12:30 <ghnatiuk> ultimately that's where filtering will happen, I think

12:30 <aslakhellesoy> ghnatiuk: it could be in cucumber too

12:30 <ghnatiuk> gherkin's fast enough that i think it will (down the road) make sense to build the entire ast, then throw parts of it out

12:30 <aslakhellesoy> -but i imagine other gherkin bases tools (SpecFlow) would want filtering too

12:31 <aslakhellesoy> so if we keep it in gherkin it's easier to port

12:32 <ghnatiuk> welllll

12:32 <ghnatiuk> seems to me like the idea of tagged/filtered things is pretty abstract

12:33 <ghnatiuk> gherkin would have to get a whole lot smarter to know what tags applied to

12:33 <aslakhellesoy> we already have gherkin/tools/pretty_listener.rb

12:33 <aslakhellesoy> I'm thinking gherkin/tools/filter_listener.rb

12:34 <aslakhellesoy> on :eof it replays events, just filtered

12:36 <* ghnatiuk> thinking

12:37 <ghnatiuk> that would mean implementing all the fancy tag logic in gherkin, right?

12:39 <aslakhellesoy> ghnatiuk: yes, but that's pretty simple now: http://github.com/aslakhellesoy/cucumber/blob/master/lib/cucumber/tag_expression.rb

12:40 <aslakhellesoy> the tag expr is converted to booleans and evaled :-D

12:40 <ghnatiuk> aslakhellesoy: That got cleaned up!

12:40 <ghnatiuk> nice!

12:42 <aslakhellesoy> http://github.com/aslakhellesoy/cucumber/blob/master/spec/cucumber/tag_expression_spec.rb

12:44 <ghnatiuk> aslakhellesoy: I really want to be able to ultimately do FeatureSuite.filter ...

12:44 <aslakhellesoy> ghnatiuk: how do you mean?

12:44 <ghnatiuk> it doesn't feel right to me, ATM, for gherkin to be concerned with that

12:44 <ghnatiuk> well, ast_node.filter

12:45 <ghnatiuk> and get back a new filtered ast

12:45 <aslakhellesoy> ghnatiuk: well. imagine you have 200 feature files

12:45 <aslakhellesoy> and you want an ast for only the stuff that has @foo

12:45 <aslakhellesoy> you'd have to build a huge ast and filter that

12:45 <ghnatiuk> that's like 2 seconds in gherkin, right :)

12:46 <aslakhellesoy> ghnatiuk: there is also the ast building overhead which happens outside gherkin

12:46 <aslakhellesoy> but it might not be a whole lot...

12:46 <ghnatiuk> hah, strangely, when running the benchmarks, treetop w/o ast building takes more time than w/ ast building

12:47 <ghnatiuk> or maybe it's building it in both cases...er

12:47 <aslakhellesoy> any particular reason you want to build the full ast first and then filter it - as opposed to filter it while it's being built?

12:47 <ghnatiuk> well, you can filter for other things than just tags

12:48 <aslakhellesoy> yes, names based on regexp and line numbers for a particular file

12:48 <aslakhellesoy> i mean - the end result is the same - a subset of the full asts

12:49 <aslakhellesoy> how we get there doesn't seem important as far as usage goes

12:49 <aslakhellesoy> i think it would be fairly troublesome to add back filtering inside the current cucumber ast

12:50 <aslakhellesoy> it's already crufty as hell, and adding that in addition would be a nightmare

12:50 <aslakhellesoy> so in order to finish the task i've started on i think it's simplest to go for filtering while building the ast

12:51 <ghnatiuk> okay

12:51 <msassak> aslakhellesoy ghnatiuk what's the simplest thing that will work, right now?

12:52 <aslakhellesoy> right now, adding filtering to the ast seems too hard, so filtering while building the ast seems simpler

12:52 <ghnatiuk> Treetop already does it. Making gherkin do it too so it can be dropped in makes sense.

12:52 <aslakhellesoy> treetop does it by looking ahead

12:53 <msassak> i'm not sure it belongs in tools

12:53 <aslakhellesoy> and we can't look ahead with gherkin

12:53 <msassak> (speaking of adding it into gherkin)

12:53 <msassak> but i think it would be a nice addition

12:53 <msassak> the idea of chaining listeners is an attractive one, to me

12:53 <aslakhellesoy> if we add it to gherkin i think it's a matter of moving cucumber's TagExpression and a special listener

12:54 <aslakhellesoy> i think the listener will be fairly simple

12:55 <aslakhellesoy> anyone up for pairing for an hour?

12:56 <msassak> if i had my mac at work, sure, as it is... maybe

12:57 <ghnatiuk> yeah, at day job currently :)

13:00 <aslakhellesoy> ok, i'll get started then :-)

13:13 <GitHub53> gherkin: 07filter-listener 03aslakhellesoy * fd3cf9d (0 files in 0 dirs): Start work on filter listener - http://bit.ly/cjR1d2

13:14 <aslakhellesoy> ghnatiuk: msassak there you go - that's the idea.

13:14 <aslakhellesoy> of course, :eof will have to be a little smarter

13:15 <aslakhellesoy> our pretty listener makes it a tad easier for us to write specs too

13:16 <msassak> that's pretty much what i imagined

13:19 <GitHub39> gherkin: 07filter-listener 03aslakhellesoy * a7c4700 (1 files in 1 dirs): Make a method that does the hard work - http://bit.ly/cbXFMp

13:22 <nofxx> noon, allow me a lil question, does "when I select" have a inverse op? Wanna deselect an item from a multiselect.

13:31 <aslakhellesoy> nofxx: select is for <select> boxes

13:32 <aslakhellesoy> are you trying to uncheck an <input type="checkbox"> ?

13:34 <aslakhellesoy> nofxx: aaah you're using a <select multiple="multiple">?

13:37 <aslakhellesoy> nofxx: Cucumber-Rails doesn't have that, but it would be easy to add:

13:37 <aslakhellesoy> http://github.com/aslakhellesoy/cucumber-rails/blob/master/templates/skeleton/step_definitions/capybara_steps.rb.erb#L64

13:38 <aslakhellesoy> http://github.com/aslakhellesoy/cucumber-rails/blob/master/templates/skeleton/step_definitions/webrat_steps.rb.erb#L54

13:38 <aslakhellesoy> i'd be happy to merge it in if you change it

15:04 <flov> arg! Can anybody tell me why i get this exception???

15:04 <flov> http://pastie.org/874457

15:05 <flov> the given page can be seen at http://giveawaythings.org/signup

15:08 <ghnatiuk> flov: it looks like Country is not actually a label for the country select field

15:08 <ghnatiuk> so webrat doesn't know how to find the select box

15:09 <flov> ghnatiuk: why? the id of the select field is the same as the for attribute of the label tag

15:11 <ghnatiuk> hmm. source looks like (~ line 108) it's not a label

15:12 <ghnatiuk> but I didn't look super hard. Have you tried using "Then show me the page" before that step?

15:12 <ghnatiuk> often helps to see exactly what webrat is looking at

15:13 <flov> yes its passing

15:13 <flov> also i dont get an ElementNotFound exception but instead tells me that its examining a nil object

15:13 <flov> btw: i use capybara

15:13 <flov> its passing without the I Select...

15:14 <flov> oh sorry

15:15 <flov> the given code on the domain is not the current one

15:15 <ghnatiuk> hah

15:15 <flov> i was wondering :)

15:15 <ghnatiuk> are those countries in a table that needs to be seeded for the cucumber environment?

15:15 <flov> oh yes! Thats it. they are not in the db :)

15:16 <flov> ghnatiuk: Thank you, i appreciate your help

15:16 <ghnatiuk> flov: np.

15:26 <aslakhellesoy> ghnatiuk, could you do me a favour and answer Michał Nowotarski on the ML?

15:27 <aslakhellesoy> I'm going to be busy the next days.

15:29 <* ghnatiuk> reading

15:30 <nkabbara> Hi, I'm trying to test a metal using cucumber, but it seems stubbing isn't extending all the way to the metal code. Any idea why? Thanks.

15:37 <ghnatiuk> aslakhellesoy: I don't see a lot of advantage to the Path change

15:37 <ghnatiuk> PageElement seems like an interesting idea for cucumber-rails

15:46 <ghnatiuk> aslakhellesoy: Sure, I'll respond. gonna get out of the office first, though. way too nice outside to be away from windows.

15:58 <ghnatiuk> gtg

16:44 <flov> \join #rubyonrails

20:29 <bmcmanus> so i've been messing with java for a few months but am starting a new rails app... what'd i miss with this whole webrat/capybara thing?

21:04 <bmcmanus> trying to get cucumber working with rspec in rails3 and i'm getting "undefined method `have_content' for #<Cucumber::Rails::World:0x2071db0> (NoMethodError)"

21:05 <bmcmanus> anyone know why it wouldn't recognize have_content?