BDD frameworks in .NET
115 views
Skip to first unread message
David Starr
Apr 14, 2009, 11:32:10 PM4/14/09
to Behaviour Driven Development
OK, I just want to use something that isn't a dead project and helps
me find a good shared language with my customer. I have been using my
own name styling and extnesion methods for clean asserts for awhile.
Does anyone have a large regression suite in any of these? Does anyone
have experience woth more than one? I really want to hear
recomendations from others experience so I can avoid your pain. Help a
brutha out?
NBehave
StoryQ
SpecUnit.NET
More?...
me find a good shared language with my customer. I have been using my
own name styling and extnesion methods for clean asserts for awhile.
Does anyone have a large regression suite in any of these? Does anyone
have experience woth more than one? I really want to hear
recomendations from others experience so I can avoid your pain. Help a
brutha out?
NBehave
StoryQ
SpecUnit.NET
More?...
Jimmy Bogard
Apr 14, 2009, 11:35:58 PM4/14/09
to behaviordriv...@googlegroups.com
Don't forget about MSpec!
David Starr
Apr 14, 2009, 11:46:18 PM4/14/09
to behaviordriv...@googlegroups.com
I actually read Aaron's post on it (again) earlier tonight. That's the thing, this is worse than picking an IoC container :)
C'mon, Jimmy, what are you using?
Jimmy Bogard
Apr 14, 2009, 11:49:52 PM4/14/09
to behaviordriv...@googlegroups.com
Well, NBehave, naturally! It's been doing what we need for awhile, so we just make trunk commits as needed.
David Starr
Apr 14, 2009, 11:53:09 PM4/14/09
to behaviordriv...@googlegroups.com
OK, so you just aren't updating the downloads? The downloads are stale, but code is fresh? Right on! I just saw that my ElegantCode tribe member Chris Bilson is a contrib on MSpec, though. Might have to go there, too.
Raymond Lewallen
Apr 15, 2009, 12:34:44 AM4/15/09
to behaviordriv...@googlegroups.com
David, you said you are looking for getting a good shared language with your customer. That's a cognitive practice that a framework can't help you with. Your efforts are best geared towards gaining understanding, then choosing a dsl to help your organize that understanding in your code, once you are able to articulate it accurately and effectively. A dsl framework cannot help you with that. Spend a good 5 or 6 months practicing, then choose a dsl later that fits your newfound knowledge.
Ray Lewallen
Morgan Persson
Apr 15, 2009, 1:59:07 AM4/15/09
to behaviordriv...@googlegroups.com
We haven't updated the downloads in a long time, we should...
It's a pity that http://teamcity.codebetter.com/project.html?projectId=project30 requires a login, the latest build is available there.
/Morgan
It's a pity that http://teamcity.codebetter.com/project.html?projectId=project30 requires a login, the latest build is available there.
/Morgan
James Gregory
Apr 15, 2009, 5:03:35 AM4/15/09
to behaviordriv...@googlegroups.com
TeamCity has a guestAuth switch that you can add to a URL (as the
first dir after the domain) that let's you browse as the guest user.
I'm on my phone right now so I can't check if it'd work for your URL
but it definitely works for the file urls, I use them for the fluent
nh build downloads.
first dir after the domain) that let's you browse as the guest user.
I'm on my phone right now so I can't check if it'd work for your URL
but it definitely works for the file urls, I use them for the fluent
nh build downloads.
Ben Lovell
Apr 15, 2009, 5:21:16 AM4/15/09
to behaviordriv...@googlegroups.com
On Wed, Apr 15, 2009 at 5:34 AM, Raymond Lewallen <rlew...@gmail.com> wrote:
Amen to this. It seems most people coming into BDD focus on the tools rather than the practice far too early. Wise words.David, you said you are looking for getting a good shared language with your customer. That's a cognitive practice that a framework can't help you with. Your efforts are best geared towards gaining understanding, then choosing a dsl to help your organize that understanding in your code, once you are able to articulate it accurately and effectively. A dsl framework cannot help you with that. Spend a good 5 or 6 months practicing, then choose a dsl later that fits your newfound knowledge.
Ray Lewallen
Ben
David Starr
Apr 15, 2009, 10:05:48 AM4/15/09
to behaviordriv...@googlegroups.com
I hear and get that, guys. I have been using the structure and discussion for awhile. I probably wasn't careful enough when describing things above.
What I am really after here is tooling to support the existing discussion.
David Starr
elegantcode.com | pluralsight.com
What I am really after here is tooling to support the existing discussion.
David Starr
elegantcode.com | pluralsight.com
David Starr
Apr 15, 2009, 10:13:35 AM4/15/09
to behaviordriv...@googlegroups.com
Jimmy, I can get to the artifacts on the CI server via guest login. Thanks for the pointer.
David Starr
elegantcode.com | pluralsight.com
David Starr
elegantcode.com | pluralsight.com
Brandon
Apr 26, 2009, 4:37:37 PM4/26/09
to Behaviour Driven Development
I'm also looking for a good BDD tool for C#.
nBehave looks good, but I can't find any documentation?
SpecUnit.NET is in the lead for me right now.
I just looked at specter, which looks interesting, but I'm not so
sure. I don't know how well that'd work in Visual Studio?
I just don't like the syntax of MSpec.
The thing with nBehave and SpecUnit.NET is that I hate all those
underscores. C# has the convention of using camel case, and I'd just
rather stick with that.
I'm almost to the point of just using nUnit and having a naming
convention for the tests.
nBehave looks good, but I can't find any documentation?
SpecUnit.NET is in the lead for me right now.
I just looked at specter, which looks interesting, but I'm not so
sure. I don't know how well that'd work in Visual Studio?
I just don't like the syntax of MSpec.
The thing with nBehave and SpecUnit.NET is that I hate all those
underscores. C# has the convention of using camel case, and I'd just
rather stick with that.
I'm almost to the point of just using nUnit and having a naming
convention for the tests.
Dan North
Apr 26, 2009, 8:55:19 PM4/26/09
to behaviordriv...@googlegroups.com
Hi David.
I've been watching this thread with interest. You ask a reasonable question, and I'm not sure I have an answer for you.
The problem I am seeing in the .NET space is that people keep producing Little Languages or frameworks just because they can - not necessarily because the framework is useful - or because they want to demonstrate their dot-net-fu.
JBehave and cucumber (the rspec scenario runner) work because of what they don't have. They simply map a line from a text file (starting with "Given", "When" or "Then") to an annotated method or block of code, possibly with parameters. That's it. That's all they do. In fact they both use other frameworks (JUnit and RSpec, respectively) to execute the scenarios.
BDD works at two levels. At the outermost level it is analagous to automated acceptance testing - each scenario represents an acceptance test. It is quite likely you will want to reuse steps from one acceptance test in another. In fact the point of scenarios is to perform the same event (the When step) in different contexts (different combinations of Givens) and to expect different outcomes (the Thens). This is where a BDD framework can be useful.
At the code level BDD is coding-by-example (aka TDD), where you don't care so much about reuse of steps. Each example (or unit test in TDD-speak) demonstrates a specific aspect of the behaviour of an object, or its interactions with other objects, at a detail level.
It seems that all the .NET "BDD frameworks" are focused on this lower level - where tools like NUnit are already perfectly adequate and where they are almost always overkill. This means there isn't anything analagous to JBehave or cucumber - a plain text scenario runner with reusable steps - in .NET.
I would love to see NBehave become a port of JBehave but taking advantage of C# semantics and language features. Right now there is a bunch of fragmentation in the .NET space. I haven't even heard of half these frameworks, which is a shame because I'd be happy to correspond with their authors. I have shared this vision with Morgan Persson and Joe Ocampo (the NBehave guys) and I am confident they are going in the right direction, but I haven't been in touch with any of the others.
Whatever they are promoting it isn't anything I would call BDD. The point of BDD is to bring business stakeholders, analysts, testers and programmers together to describe application behaviour in a consistent and useful way. This splintering of frameworks, syntax-fu and not-invented-here syndrome is the antithesis of that.
It would be great to see some cohesion across the various frameworks to create a robust BDD scenario runner for .NET. That would make me very happy.
Cheers,
Dan
ps. In the meantime I would keep an eye on cucumber on iron ruby. It isn't very mature yet but some people are trying it already.
2009/4/27 Brandon <program...@gmail.com>
I've been watching this thread with interest. You ask a reasonable question, and I'm not sure I have an answer for you.
The problem I am seeing in the .NET space is that people keep producing Little Languages or frameworks just because they can - not necessarily because the framework is useful - or because they want to demonstrate their dot-net-fu.
JBehave and cucumber (the rspec scenario runner) work because of what they don't have. They simply map a line from a text file (starting with "Given", "When" or "Then") to an annotated method or block of code, possibly with parameters. That's it. That's all they do. In fact they both use other frameworks (JUnit and RSpec, respectively) to execute the scenarios.
BDD works at two levels. At the outermost level it is analagous to automated acceptance testing - each scenario represents an acceptance test. It is quite likely you will want to reuse steps from one acceptance test in another. In fact the point of scenarios is to perform the same event (the When step) in different contexts (different combinations of Givens) and to expect different outcomes (the Thens). This is where a BDD framework can be useful.
At the code level BDD is coding-by-example (aka TDD), where you don't care so much about reuse of steps. Each example (or unit test in TDD-speak) demonstrates a specific aspect of the behaviour of an object, or its interactions with other objects, at a detail level.
It seems that all the .NET "BDD frameworks" are focused on this lower level - where tools like NUnit are already perfectly adequate and where they are almost always overkill. This means there isn't anything analagous to JBehave or cucumber - a plain text scenario runner with reusable steps - in .NET.
I would love to see NBehave become a port of JBehave but taking advantage of C# semantics and language features. Right now there is a bunch of fragmentation in the .NET space. I haven't even heard of half these frameworks, which is a shame because I'd be happy to correspond with their authors. I have shared this vision with Morgan Persson and Joe Ocampo (the NBehave guys) and I am confident they are going in the right direction, but I haven't been in touch with any of the others.
Whatever they are promoting it isn't anything I would call BDD. The point of BDD is to bring business stakeholders, analysts, testers and programmers together to describe application behaviour in a consistent and useful way. This splintering of frameworks, syntax-fu and not-invented-here syndrome is the antithesis of that.
It would be great to see some cohesion across the various frameworks to create a robust BDD scenario runner for .NET. That would make me very happy.
Cheers,
Dan
ps. In the meantime I would keep an eye on cucumber on iron ruby. It isn't very mature yet but some people are trying it already.
2009/4/27 Brandon <program...@gmail.com>
Thomas Lundström
Apr 27, 2009, 1:14:44 AM4/27/09
to behaviordriv...@googlegroups.com
Hi,
Regarding Cucumber on IronRuby; it's very immature as of now. The
latest version of Cucumber doesn't run on IronRuby 0.3 (or 0.2 for
that matter).
If you are interested to try it out, I recently got Cucumber 0.1.16
running together with rspec 1.1.12 (IIRC) on IronRuby 0.3, a setup
which I'll demo speccing ASPNET MVC controllers at the ALT.NET Oresund
meeting in Malmö, Sweden, today. I plan to blog about it later this
week.
Best regards,
Thomas Lundström
http://thomaslundstrom.blogspot.com
Regarding Cucumber on IronRuby; it's very immature as of now. The
latest version of Cucumber doesn't run on IronRuby 0.3 (or 0.2 for
that matter).
If you are interested to try it out, I recently got Cucumber 0.1.16
running together with rspec 1.1.12 (IIRC) on IronRuby 0.3, a setup
which I'll demo speccing ASPNET MVC controllers at the ALT.NET Oresund
meeting in Malmö, Sweden, today. I plan to blog about it later this
week.
Best regards,
Thomas Lundström
http://thomaslundstrom.blogspot.com
Morgan Persson
Apr 27, 2009, 2:59:51 AM4/27/09
to behaviordriv...@googlegroups.com
Hi
just a note, NBehave has support for mapping a line of text in a text file to an annotated method (method with attribute in .net), its not released yet and currently supports rather simple stuff. For example, you cant do JBehabe's tic-tac-toe example, but I'm working on it.
The easiest way to check out this feature is by looking at the source code.
http://code.google.com/p/nbehave/source/browse/trunk/src/NBehave.Framework.Tests/GreetingSystem.txt
and
http://code.google.com/p/nbehave/source/browse/trunk/src/TestPlainTextAssembly/GreetingSystemActionSteps.cs
Then you can run it from the command line with "nbehave-console.exe TestPlainTextAssembly.dll /sf=GreetingSystem.txt (if you grab the fix I just checked in ;)
No support for this feature in msbuild or nant yet. (Anyone want to send me a patch for this?)
/Morgan
just a note, NBehave has support for mapping a line of text in a text file to an annotated method (method with attribute in .net), its not released yet and currently supports rather simple stuff. For example, you cant do JBehabe's tic-tac-toe example, but I'm working on it.
The easiest way to check out this feature is by looking at the source code.
http://code.google.com/p/nbehave/source/browse/trunk/src/NBehave.Framework.Tests/GreetingSystem.txt
and
http://code.google.com/p/nbehave/source/browse/trunk/src/TestPlainTextAssembly/GreetingSystemActionSteps.cs
Then you can run it from the command line with "nbehave-console.exe TestPlainTextAssembly.dll /sf=GreetingSystem.txt (if you grab the fix I just checked in ;)
No support for this feature in msbuild or nant yet. (Anyone want to send me a patch for this?)
/Morgan
On Mon, Apr 27, 2009 at 2:55 AM, Dan North <tast...@gmail.com> wrote:
Mark Ryall
Apr 27, 2009, 9:05:22 AM4/27/09
to Behaviour Driven Development
I've worked on a few projects where the technology used for system
specification was not the same as that used to implement the system
under test. It's an excellent application of a dynamic language.
When building a web application (.net, java or otherwise) my
preference would be to use ruby: cucumber + webrat + mechanize/
selenium. The act of using different language makes the context shift
between reading your application code and specification code much
cleaner - just as you know what part of the application you're dealing
with when looking at some SQL or javascript.
It also means that there's a lower barrier to entry for QAs and BAs to
actually writing the features and scenarios as they don't need to be
overwhelmed by an IDE.
Testing a wpf or winforms application is a bit more challenging
although it should (theoretically) be possible to build an ironruby
wrapper around something like white.
Clearly you need to use .net tools for unit and integration testing
but an external DSL such as cucumber makes the most sense to me for
specifying overall system behaviour.
Mark.
On Apr 27, 10:55 am, Dan North <tasta...@gmail.com> wrote:
> Hi David.
>
> I've been watching this thread with interest. You ask a reasonable question,
> and I'm not sure I have an answer for you.
>
> The problem I am seeing in the .NET space is that people keep producing
> Little Languages or frameworks just because they can - not necessarily
> because the framework is useful - or because they want to demonstrate their
> dot-net-fu.
>
> JBehave and cucumber (the rspec scenario runner) work because of what they *
> don't* have. They simply map a line from a text file (starting with "Given",
> it<http://weblogs.asp.net/astopford/archive/2009/04/14/bdd-with-rspec-cu...>already.
>
> 2009/4/27 Brandon <programmer2...@gmail.com>
specification was not the same as that used to implement the system
under test. It's an excellent application of a dynamic language.
When building a web application (.net, java or otherwise) my
preference would be to use ruby: cucumber + webrat + mechanize/
selenium. The act of using different language makes the context shift
between reading your application code and specification code much
cleaner - just as you know what part of the application you're dealing
with when looking at some SQL or javascript.
It also means that there's a lower barrier to entry for QAs and BAs to
actually writing the features and scenarios as they don't need to be
overwhelmed by an IDE.
Testing a wpf or winforms application is a bit more challenging
although it should (theoretically) be possible to build an ironruby
wrapper around something like white.
Clearly you need to use .net tools for unit and integration testing
but an external DSL such as cucumber makes the most sense to me for
specifying overall system behaviour.
Mark.
On Apr 27, 10:55 am, Dan North <tasta...@gmail.com> wrote:
> Hi David.
>
> I've been watching this thread with interest. You ask a reasonable question,
> and I'm not sure I have an answer for you.
>
> The problem I am seeing in the .NET space is that people keep producing
> Little Languages or frameworks just because they can - not necessarily
> because the framework is useful - or because they want to demonstrate their
> dot-net-fu.
>
> don't* have. They simply map a line from a text file (starting with "Given",
>
> 2009/4/27 Brandon <programmer2...@gmail.com>
Reply all
Reply to author
Forward
0 new messages