On Jan 29, 2007, at 10:58 PM, Al Byers wrote:

Could this be a matter of differing requirements? I am guessing that one
could think of testing as a means to insure that the core code is
functioning properly. I realize that I was thinking about what I would want
from a project management point of view.

In that case, I think my requirements would be:

1. Something that would let me jump around and assess the state of multiple
projects. That would imply that I do not want to use a command-line
interface. I don't want to have to log in to different systems. I would
want to see the available tests and pick all or some of them.

2. If I want to check the state of a task, I want to be able to run a test that is appropriate for that task whether it be an acceptance test on the response to a request or an internal test on business logic functionality. I would like the same command interface for all tests. Of course, WebTest does
not have such an interface, but it does not seem like it would be too
difficult to glean the available tests and descriptions from the ant script
files.


It's way easy to do this in ant - I have a bunch of examples for this somewhere in my build files. I can pull it together whenever you're ready Al. It doesn't prompt you though (but it could still be VERY easy to control with arguments being passed in)

I see WebTest as a vehicle for doing the above. It is designed for web-based tests, but it seems like with a little work it could also be a harness for running JUnit tests. Perhaps a special event handler for running JUnit tests
and another for wrapping web requests with a transaction.

Let me try again at refining my wish list:

1. A prompting interface.
2. The ability to run all tests from that interface.

-Al

On 1/29/07, David E. Jones <[EMAIL PROTECTED]> wrote:


I'd really prefer to do what has been proposed as a best practice and
write tests using the same OFBiz framework tools that we use to write
applications, like simple-methods, services, etc...

But yes, it is possible to call a service through a web request and
there is one in the webtools wecapp that has been there for years.
The trick is you have to set export="true" for all services called
this way, which is another reason to do logic-level test (including
service calls) in a more black-box way, especially if they are not
for testing things that are intended to be available externally.

-David


On Jan 29, 2007, at 4:01 AM, Andrew Sykes wrote:

> Assuming an automated web browser type technology is the way to go for
> testing...
>
> What does everyone think of having an option to run a service
> synchronously from webtools?
>
> This would allow a lot of clever asserts from the test tool?
> Without the
> need to make the tool dispatcher aware? Would this be an adequate
> approach?
> --
> Kind Regards
> Andrew Sykes <[EMAIL PROTECTED]>
> Sykes Development Ltd
> http://www.sykesdevelopment.com
>





Attachment: smime.p7s
Description: S/MIME cryptographic signature

Reply via email to