RESULTS-STATUS tells you if there are any non-"TEST-PASSED" results. But it doesn't provide any way for you to get your hands on these results. Originally, I thought that one could simply test for them, but the results categories aren't actually exported from FiveAM, so this is difficult.
To make it easier to find "interesting" results, I modified RESULTS-STATUS to return a second value. The first value is a boolean, as before, telling you if the tests all passed. If they didn't, the second value is now a list of the non-passing results.
I believe this to be a benign change from the standpoint of backward compatibility, since anyone just looking for a single return value will get the same boolean value as before.
It's possible that the actual value --- not interpreted as a boolean --- will be different, since the new code uses REMOVE-IF instead of EVERY. But if a caller takes information out of the results list in a function whose documentation proclaims it to be boolean, I submit that the caller is doing so at his/her own risk!
Proposed patch is attached.
Best, Robert
On Mon, 2012-09-10 at 13:53 -0500, Robert Goldman wrote:
RESULTS-STATUS tells you if there are any non-"TEST-PASSED" results. But it doesn't provide any way for you to get your hands on these results. Originally, I thought that one could simply test for them, but the results categories aren't actually exported from FiveAM, so this is difficult.
To make it easier to find "interesting" results, I modified RESULTS-STATUS to return a second value. The first value is a boolean, as before, telling you if the tests all passed. If they didn't, the second value is now a list of the non-passing results.
I believe this to be a benign change from the standpoint of backward compatibility, since anyone just looking for a single return value will get the same boolean value as before.
It's possible that the actual value --- not interpreted as a boolean --- will be different, since the new code uses REMOVE-IF instead of EVERY. But if a caller takes information out of the results list in a function whose documentation proclaims it to be boolean, I submit that the caller is doing so at his/her own risk!
Proposed patch is attached.
Applied with some modifications, thanks :)
On 11/17/12 Nov 17 -10:45 PM, Stelian Ionescu wrote:
On Mon, 2012-09-10 at 13:53 -0500, Robert Goldman wrote:
RESULTS-STATUS tells you if there are any non-"TEST-PASSED" results. But it doesn't provide any way for you to get your hands on these results. Originally, I thought that one could simply test for them, but the results categories aren't actually exported from FiveAM, so this is difficult.
To make it easier to find "interesting" results, I modified RESULTS-STATUS to return a second value. The first value is a boolean, as before, telling you if the tests all passed. If they didn't, the second value is now a list of the non-passing results.
I believe this to be a benign change from the standpoint of backward compatibility, since anyone just looking for a single return value will get the same boolean value as before.
It's possible that the actual value --- not interpreted as a boolean --- will be different, since the new code uses REMOVE-IF instead of EVERY. But if a caller takes information out of the results list in a function whose documentation proclaims it to be boolean, I submit that the caller is doing so at his/her own risk!
Proposed patch is attached.
Applied with some modifications, thanks :)
Question: should we bump the ASDF version lisp-expr to indicate that this return variable is available? I.e., move to version 1.1?
Not a big deal, but it's nice to stick to the convention that a change in the API merits a .1 version bump....
Cheers, r
On Sat, 2012-11-24 at 18:52 -0600, Robert Goldman wrote:
On 11/17/12 Nov 17 -10:45 PM, Stelian Ionescu wrote:
On Mon, 2012-09-10 at 13:53 -0500, Robert Goldman wrote:
RESULTS-STATUS tells you if there are any non-"TEST-PASSED" results. But it doesn't provide any way for you to get your hands on these results. Originally, I thought that one could simply test for them, but the results categories aren't actually exported from FiveAM, so this is difficult.
To make it easier to find "interesting" results, I modified RESULTS-STATUS to return a second value. The first value is a boolean, as before, telling you if the tests all passed. If they didn't, the second value is now a list of the non-passing results.
I believe this to be a benign change from the standpoint of backward compatibility, since anyone just looking for a single return value will get the same boolean value as before.
It's possible that the actual value --- not interpreted as a boolean --- will be different, since the new code uses REMOVE-IF instead of EVERY. But if a caller takes information out of the results list in a function whose documentation proclaims it to be boolean, I submit that the caller is doing so at his/her own risk!
Proposed patch is attached.
Applied with some modifications, thanks :)
Question: should we bump the ASDF version lisp-expr to indicate that this return variable is available? I.e., move to version 1.1?
Yes, I'm going to make a new release soon.
-
Robert Goldman -
Stelian Ionescu