[rspec-users] Upgrade problems/questions regarding RSpec 1.3.0 to 2.0.1
Jarmo Pertman
I'm trying to migrate my scripts from RSpec 1.3.0 to 2.0.1 having some
issues so far.
1) I don't like is that ".rspec" has to be in the working directory
and cannot be anymore in "spec". Okay, i can issue "mv spec/
spec.opts .rspec" in my projects, but why isn't it supported?
2) If there is somewhere "require 'spec'" in existing specs or
spec_helper or anywhere then test will not be run and this will be
shown instead:
Finished in 0 seconds
0 examples, 0 failures
It is also possible to reproduce it like this from command line:
rspec -rspec blah_spec.rb
3) Why does html formatter snippet show always failing line as:
raise(RSpec::Expectations::ExpectationNotMetError.new(message))
in 1.x it would show the failing line in the spec file instead... is
this a bug?
4) In 1.x i modified ExampleGroup's description in my html formatter
to have some additional information in it like timestamp. I did it in
example_group_started where the passed variable example_group was an
instance of the ExampleGroup. In 2.x this is just a class name. How
can i accomplish the similar functionality?
5) I don't understand how to use multiple formatters together. In
RSpec 1.x it is possible to use multiple formatters like this:
spec --format nested --format html:output/blah.html
In RSpec 2.x it seems to me that only the last --format is used. The
command below:
rspec -f html -f d blah_spec.rb
Uses only nested formatter. In RSpec 1.x you had to specify the file
path like this:
rspec -f html:path_to_output
But it seems that in 2.x you have to use --out for that which
overrides all the output of course. And on top of if there is
everywhere used a term "formatter" in RSpec::Core::Configuration, but
"formatters" (in plural) in RSpec::Core::Reporter. It's constructor
also allows to pass multiple formatters into it.
I'd like to use two formatters together - documentation and html.
Documentation would be helpful for faster feedback and for an
indication of end of the testrun on the command line, but (custom)
html formatter would include more information for better
troubleshooting.
I managed to do that currently by little monkey-patching and manual
coding:
module RSpec
module Core
class Configuration
def reporter=(reporter)
@reporter = reporter
end
public :built_in_formatter
end
end
end
and in spec_helper or somewhere:
config = RSpec.configuration
config.color_enabled = true
documentation_formatter =
config.built_in_formatter(:documentation).new(config.output)
@@html_formatter = MyHtmlFormatter.new("blah/index.html") # it takes
file path as a string due to archival and file in use needs
config.reporter = RSpec::Core::Reporter.new(documentation_formatter,
@@html_formatter)
Is there some easier/more normal way to do this?
Does this mean that RSpec 2 is meant to be used with only 1 formatter
simultaneously?
_______________________________________________
rspec-users mailing list
rspec...@rubyforge.org
http://rubyforge.org/mailman/listinfo/rspec-users
Jarmo Pertman
> 3) Why does html formatter snippet show always failing line as:
> raise(RSpec::Expectations::ExpectationNotMetError.new(message))
>
> in 1.x it would show the failing line in the spec file instead... is
> this a bug?
I cannot reproduce this problem currently with standard html formatter
although i don't do anything with snippet extractor in my custom html
formatter. I need to investigate why the problem happens now and then.
Jarmo Pertman
-----
IT does really matter - http://www.itreallymatters.net
Jarmo Pertman
> I cannot reproduce this problem currently with standard html formatter
> although i don't do anything with snippet extractor in my custom html
> formatter. I need to investigate why the problem happens now and then.
I've managed to reproduce it again and always like this:
# some_spec.rb
describe "something" do
it "doesn't work" do
true.should be_false
end
end
C:\Users\jarmo\Desktop\minu\projects\Ruby\blah2>rspec some_spec.rb -f
html -o out.html
And this is the output:
http://dl.dropbox.com/u/2731643/misc/out.html
with -b the whole stacktrace is as following:
c:/ruby/lib/ruby/gems/1.8/gems/rspec-expectations-2.0.1/lib/rspec/
expectations/fail_with.rb:29:in `fail_with'
c:/ruby/lib/ruby/gems/1.8/gems/rspec-expectations-2.0.1/lib/rspec/
expectations/handler.rb:21:in `handle_matcher'
c:/ruby/lib/ruby/gems/1.8/gems/rspec-expectations-2.0.1/lib/rspec/
expectations/extensions/kernel.rb:27:in `should'
./some_spec.rb:3
c:/ruby/lib/ruby/gems/1.8/gems/rspec-core-2.0.1/lib/rspec/core/
example.rb:47:in `instance_eval'
c:/ruby/lib/ruby/gems/1.8/gems/rspec-core-2.0.1/lib/rspec/core/
example.rb:47:in `run'
c:/ruby/lib/ruby/gems/1.8/gems/rspec-core-2.0.1/lib/rspec/core/
example.rb:86:in `call'
c:/ruby/lib/ruby/gems/1.8/gems/rspec-core-2.0.1/lib/rspec/core/
example.rb:86:in `with_around_hooks'
c:/ruby/lib/ruby/gems/1.8/gems/rspec-core-2.0.1/lib/rspec/core/
example.rb:44:in `run'
c:/ruby/lib/ruby/gems/1.8/gems/rspec-core-2.0.1/lib/rspec/core/
example.rb:80:in `call'
c:/ruby/lib/ruby/gems/1.8/gems/rspec-core-2.0.1/lib/rspec/core/
example.rb:80:in `with_pending_capture'
c:/ruby/lib/ruby/gems/1.8/gems/rspec-core-2.0.1/lib/rspec/core/
example.rb:79:in `catch'
c:/ruby/lib/ruby/gems/1.8/gems/rspec-core-2.0.1/lib/rspec/core/
example.rb:79:in `with_pending_capture'
c:/ruby/lib/ruby/gems/1.8/gems/rspec-core-2.0.1/lib/rspec/core/
example.rb:43:in `run'
c:/ruby/lib/ruby/gems/1.8/gems/rspec-core-2.0.1/lib/rspec/core/
example_group.rb:260:in `run_examples'
c:/ruby/lib/ruby/gems/1.8/gems/rspec-core-2.0.1/lib/rspec/core/
example_group.rb:256:in `map'
c:/ruby/lib/ruby/gems/1.8/gems/rspec-core-2.0.1/lib/rspec/core/
example_group.rb:256:in `run_examples'
c:/ruby/lib/ruby/gems/1.8/gems/rspec-core-2.0.1/lib/rspec/core/
example_group.rb:230:in `run'
c:/ruby/lib/ruby/gems/1.8/gems/rspec-core-2.0.1/lib/rspec/core/
command_line.rb:26:in `run'
c:/ruby/lib/ruby/gems/1.8/gems/rspec-core-2.0.1/lib/rspec/core/
command_line.rb:26:in `map'
c:/ruby/lib/ruby/gems/1.8/gems/rspec-core-2.0.1/lib/rspec/core/
command_line.rb:26:in `run'
c:/ruby/lib/ruby/gems/1.8/gems/rspec-core-2.0.1/lib/rspec/core/
reporter.rb:11:in `report'
c:/ruby/lib/ruby/gems/1.8/gems/rspec-core-2.0.1/lib/rspec/core/
command_line.rb:23:in `run'
c:/ruby/lib/ruby/gems/1.8/gems/rspec-core-2.0.1/lib/rspec/core/
runner.rb:55:in `run_in_process'
c:/ruby/lib/ruby/gems/1.8/gems/rspec-core-2.0.1/lib/rspec/core/
runner.rb:46:in `run'
c:/ruby/lib/ruby/gems/1.8/gems/rspec-core-2.0.1/lib/rspec/core/
runner.rb:10:in `autorun'
c:/Ruby/bin/rspec:19
Jarmo Pertman
seem to load RSpec.
Consider the following project structure:
# Rakefile
require 'rspec/core/rake_task'
RSpec::Core::RakeTask.new(:spec) do |spec|
end
RSpec::Core::RakeTask.new(:rcov) do |spec|
spec.rcov = true
end
# lib/my_lib.rb
class MyLib
end
# spec/my_lib_spec.rb
require "lib/my_lib"
describe "should work" do
it "doesn't" do
MyLib.new.should be
end
end
Now, when executing `rake spec` everything works - RSpec itself gets
loaded automatically. When executing `rake rcov` on the other hand
then this will be the output:
C:\Users\jarmo\Desktop\minu\projects\Ruby\blah>rake rcov
(in C:/Users/jarmo/Desktop/minu/projects/Ruby/blah)
c:/ruby/bin/ruby.exe -S rcov -Ispec "./spec/my_lib_spec.rb"
./spec/my_lib_spec.rb:1: undefined method `describe' for main:Object
(NoMethodError)
from c:/ruby/lib/ruby/gems/1.8/gems/rcov-0.9.8/bin/rcov:511:in
`load'
from c:/ruby/lib/ruby/gems/1.8/gems/rcov-0.9.8/bin/rcov:511
from c:/ruby/bin/rcov:19:in `load'
from c:/ruby/bin/rcov:19
rake aborted!
ruby -S rcov -Ispec "./spec/my_lib_spec.rb" failed
It will pass of course if i'll add `require "rspec"` to somewhere.
Jarmo Pertman
RSpec.configure block like this:
# spec/my_lib_spec.rb
require "rspec"
describe "should work" do
it "with rcov also" do
MyLib.should be
my_method.should == "works"
end
end
# lib/my_lib.rb
require "rspec"
module MyLib
def my_method
"works"
end
end
RSpec.configure do |config|
config.include(MyLib)
end
# .rspec
-r
lib/my_lib
and Rakefile is the same as before.
When using `rake spec` then example passes, but with `rake rcov` i see
this:
C:\Users\jarmo\Desktop\minu\projects\Ruby\blah>rake rcov
(in C:/Users/jarmo/Desktop/minu/projects/Ruby/blah)
c:/ruby/bin/ruby.exe -S rcov -Ispec "./spec/my_lib_spec.rb"
F
Failures:
1) should work with rcov also
Failure/Error: my_method.should == "works"
undefined local variable or method `my_method' for
#<RSpec::Core::ExampleGroup::Nested_1:0x59cb488 @__memoized={}>
# ./spec/my_lib_spec.rb:6
Finished in 0.014 seconds
1 example, 1 failure
rake aborted!
ruby -S rcov -Ispec "./spec/my_lib_spec.rb" failed
(See full trace by running task with --trace)
When i put a breakpoint into example block then with `rake spec` i can
see that module is included, but not with `rake rcov` although
my_lib.rb is loaded because the first expectation succeeds as seen
above.
With `rake spec`:
(rdb:1) l
[3, 12] in C:/Users/jarmo/Desktop/minu/projects/Ruby/blah/spec/
my_lib_spec.rb
3 describe "should work" do
4 it "with rcov also" do
5 MyLib.should be
6 require "ruby-debug"
7 debugger
=> 8 my_method.should == "works"
9 end
10 end
(rdb:1) self.class.included_modules
[MyLib, RSpec::Matchers, RSpec::Core::MockFrameworkAdapter,
RSpec::Core::Pending, RSpec::Core::Let::InstanceMethods,
RSpec::Core::Let, RSpec::Core::Subject,
RSpec::Core::Extensions::InstanceEvalWithArgs, RSpec::Mocks::Methods,
PP::ObjectMixin, RSpec::Core::SharedExampleGroup,
RSpec::Core::ObjectExtensions, Kernel]
With `rake rcov`:
(rdb:1) self.class.included_modules
[RSpec::Matchers, RSpec::Core::MockFrameworkAdapter,
RSpec::Core::Pending, RSpec::Core::Let::InstanceMethods,
RSpec::Core::Let, RSpec::Core::Subject,
RSpec::Core::Extensions::InstanceEvalWithArgs, RSpec::Mocks::Methods,
PP::ObjectMixin, RSpec::Core::SharedExampleGroup,
RSpec::Core::ObjectExtensions, Kernel]
So, for some reason the module is not included when running with RCov.
By the way all the specs were working before with 1.3.0.
Jarmo Pertman
-----
IT does really matter - http://www.itreallymatters.net
On Nov 4, 10:37 pm, Jarmo Pertman <jarm...@gmail.com> wrote:
> There seems to be also a problem when using RCov raketask - it doesn't
> seem to load RSpec.
Jarmo Pertman
painful.
Subject#subject behavior with "self" is also broken!
This fails in RSpec 2 with stack overflow:
describe "something" do
subject { self }
it "is ok?" do
should be_ok
end
def ok?
true
end
end
It is passing in RSpec 1.3. See the original discussion of that
feature that i myself suggested/needed
http://groups.google.com/group/rspec/browse_thread/thread/89d6dd78b8f76df7/21109b78fc61a0e6
It also seems to me that the #subject specs provided with the patch
for 1.x have been deleted also during some refactoring or something.
At least i couldn't find the specs testing that explicit functionality
from rspec-core. They still exist at 1.x repo though
https://github.com/dchelimsky/rspec/blob/master/spec/spec/example/example_methods_spec.rb
I'm really sorry if some of the problems in this thread are already
fixed in master, but i couldn't see any open bugs relating to any of
the problems described here.
Waiting for some solutions-explanations how to solve most of the
problems hopefully.
Jarmo Pertman
-----
IT does really matter - http://www.itreallymatters.net
David Chelimsky
Hey Jarmo,
Sorry for all the trouble, and thanks for all of the feedback. I'm completely consumed with getting last minute things together for The RSpec Book to go to the printer (yay), and preparing for my talk next week at RubyConf, so I'm not quite sure when I'll be able to review/respond. Hopefully sometime over the weekend, but no guarantees.
In the meantime, be sure to search the closed issues on github (unfortunately, they don't show up if you search the open issues), and please feel free to open new issues for the problems you are experiencing.
Cheers,
David
David Chelimsky
> Hello!
Hello!
> I'm trying to migrate my scripts from RSpec 1.3.0 to 2.0.1 having some
> issues so far.
>
> 1) I don't like is that ".rspec" has to be in the working directory
> and cannot be anymore in "spec". Okay, i can issue "mv spec/
> spec.opts .rspec" in my projects, but why isn't it supported?
It's more common across other libraries to put "dot files" in your home and/or project root directories.
> 2) If there is somewhere "require 'spec'" in existing specs or
> spec_helper or anywhere then test will not be run and this will be
> shown instead:
> Finished in 0 seconds
> 0 examples, 0 failures
>
> It is also possible to reproduce it like this from command line:
> rspec -rspec blah_spec.rb
Sounds like a new bug. Please report to http://github.com/rspec/rspec-core/issues.
> 3) Why does html formatter snippet show always failing line as:
> raise(RSpec::Expectations::ExpectationNotMetError.new(message))
>
> in 1.x it would show the failing line in the spec file instead... is
> this a bug?
Yes. Please report to http://github.com/rspec/rspec-core/issues.
> 4) In 1.x i modified ExampleGroup's description in my html formatter
> to have some additional information in it like timestamp. I did it in
> example_group_started where the passed variable example_group was an
> instance of the ExampleGroup. In 2.x this is just a class name. How
> can i accomplish the similar functionality?
You can write a custom subclass of the HtmlFormatter and add whatever you like.
> 5) I don't understand how to use multiple formatters together.
You can't yet. rspec-core-2 was built from Micronaut, so a lot of the functionality had to be rebuilt. This is one feature I didn't forward port from rspec-1 because I didn't see the value in it for the complexity. RSpec-2 has been in the works for over a year now and you're the first person to ask about this, so it's obviously not a widely used feature, and I'm trying to keep unused features out.
Feel free to report this to http://github.com/rspec/rspec-core/issues and we can discuss it further in the issue.
David Chelimsky
David Chelimsky
> Upgrade to RSpec 2.x has been more painful than i expected. A lot more
> painful.
>
> Subject#subject behavior with "self" is also broken!
>
> This fails in RSpec 2 with stack overflow:
> describe "something" do
> subject { self }
>
> it "is ok?" do
> should be_ok
> end
>
> def ok?
> true
> end
> end
>
> It is passing in RSpec 1.3. See the original discussion of that
> feature that i myself suggested/needed
> http://groups.google.com/group/rspec/browse_thread/thread/89d6dd78b8f76df7/21109b78fc61a0e6
>
> It also seems to me that the #subject specs provided with the patch
> for 1.x have been deleted also during some refactoring or something.
rspec-core-2 was a complete rewrite, so it's not that specs or code were deleted, just not all were forward ported.
> At least i couldn't find the specs testing that explicit functionality
> from rspec-core. They still exist at 1.x repo though
> https://github.com/dchelimsky/rspec/blob/master/spec/spec/example/example_methods_spec.rb
These were largely about a module re-opening issue. I added a github issue for that: https://github.com/rspec/rspec-core/issues/issue/210 but I'm not sure it solves your need. If you don't think it will, please report this one as well to https://github.com/rspec/rspec-core/issues.
> I'm really sorry if some of the problems in this thread are already
> fixed in master, but i couldn't see any open bugs relating to any of
> the problems described here.
I don't think any of these issues have been reported. Most of the problems you're encountering have to do with unconventional use, which is why they haven't come up before.
Thanks again for all the feedback.
Cheers,
David
> Waiting for some solutions-explanations how to solve most of the
> problems hopefully.
_______________________________________________
David Chelimsky
Jarmo Pertman
core before opening up any new issues. The main blockers for me to
start using RSpec 2 are currently the subject {self} and html
formatter snippet problems.
Hopefully you'll find some time to answer to my provided feedback.
Jarmo Pertman
-----
IT does really matter - http://www.itreallymatters.net
David Chelimsky
> I'd wait for some preliminary answers from you or anyone else from
> core before opening up any new issues. The main blockers for me to
> start using RSpec 2 are currently the subject {self} and html
> formatter snippet problems.
>
> Hopefully you'll find some time to answer to my provided feedback.
I'd really prefer you just submit the issues to http://github.com/rspec/rspec-core/issues. Keeps the conversation about each issue separate and ensures they each stay on the radar and get addressed, even if not immediately.
Thanks,
David
Jarmo Pertman
that you had already answered.
Going to create those issues soon.
Thanks for your feedback!
Jarmo
On Nov 5, 5:06 pm, David Chelimsky <dchelim...@gmail.com> wrote:
> On Nov 5, 2010, at 7:02 AM, Jarmo Pertman wrote:
>
> > I'd wait for some preliminary answers from you or anyone else from
> > core before opening up any new issues. The main blockers for me to
> > start using RSpec 2 are currently the subject {self} and html
> > formatter snippet problems.
>
> > Hopefully you'll find some time to answer to my provided feedback.
>
> I'd really prefer you just submit the issues tohttp://github.com/rspec/rspec-core/issues. Keeps the conversation about each issue separate and ensures they each stay on the radar and get addressed, even if not immediately.
>
> Thanks,
> David
>
>
>
>
>
>
>
>
>
>
>
> > Jarmo Pertman
> > -----
> > IT does really matter -http://www.itreallymatters.net
>
> > On Nov 5, 5:11 am, David Chelimsky <dchelim...@gmail.com> wrote:
> >> Hey Jarmo,
>
> >> Sorry for all the trouble, and thanks for all of the feedback. I'm completely consumed with getting last minute things together for The RSpec Book to go to the printer (yay), and preparing for my talk next week at RubyConf, so I'm not quite sure when I'll be able to review/respond. Hopefully sometime over the weekend, but no guarantees.
>
> >> In the meantime, be sure to search the closed issues on github (unfortunately, they don't show up if you search the open issues), and please feel free to open new issues for the problems you are experiencing.
>
> >> Cheers,
> >> David
> > _______________________________________________
> > rspec-users mailing list
> >http://rubyforge.org/mailman/listinfo/rspec-users
>
> _______________________________________________
> rspec-users mailing list
> rspec-us...@rubyforge.orghttp://rubyforge.org/mailman/listinfo/rspec-users
Jarmo Pertman
> > 4) In 1.x i modified ExampleGroup's description in my html formatter
> > to have some additional information in it like timestamp. I did it in
> > example_group_started where the passed variable example_group was an
> > instance of the ExampleGroup. In 2.x this is just a class name. How
> > can i accomplish the similar functionality?
>
> You can write a custom subclass of the HtmlFormatter and add whatever you like.
Since i don't consider this as an issue to be reported at GitHub
rather than a new way to do things then i just want to know how should
i do this. I understand that i can write a custom subclass of the
HtmlFormatter and do there whatever i like. This is the actual case -
i already have an subclass of HtmlFormatter, but the problem is that
super class is acting differently than it was in 1.x.
In HtmlFormatter#example_group_started a parameter called
example_group was an actual instance of the current ExampleGroup in
1.x thus i were able to modify it's description so it would have
additional information in my custom html formatter report. In 2.x the
parameter example_group is not anymore an instance of an ExampleGroup,
but just a class constant for runtime ExampleGroup class, thus there's
nothing useful to do with that parameter anymore. How should/could i
modify the description now? I'm afraid that in
HtmlFormatter#example_started where i could access the ExampleGroup
from the Example instance it's already too late since ExampleGroup
description might be already flushed to the report. Haven't tried that
yet though.
Hopefully you understand my concern better now.
Jarmo Pertman
Jarmo Pertman
overview of the outcome of this thread.
1) 0 examples will be run if requiring "spec"
https://github.com/rspec/rspec-core/issues/issue/212
2) add possibility to use multiple formatters at the same time
https://github.com/rspec/rspec-core/issues/issue/213
3) using HtmlFormatter shows RSpec's internal snippet in the report
https://github.com/rspec/rspec-core/issues/issue/214
4) RCov RakeTask doesn't load RSpec
https://github.com/rspec/rspec-core/issues/issue/215
5) using RCov RakeTask doesn't include modules into example groups
with RSpec.configure
https://github.com/rspec/rspec-core/issues/issue/216
6) Subject#subject {self} causes stack overflow:
https://github.com/rspec/rspec-core/issues/issue/217
Jarmo Pertman
-----
IT does really matter - http://www.itreallymatters.net
On Nov 5, 5:06 pm, David Chelimsky <dchelim...@gmail.com> wrote:
> I'd really prefer you just submit the issues tohttp://github.com/rspec/rspec-core/issues. Keeps the conversation about each issue separate and ensures they each stay on the radar and get addressed, even if not immediately.
>
> Thanks,
> David