Hi Lukas,

Did you receive this? I got an error message from the mailing list.

The list's address changed to pharo-project@lists.gforge.inria.fr for some reason...

You asked me ...

Thanks for your feedback!

...

Lukas you should have a look no super read different super black simple super green

I do not understand, so I am reading <http://smallwiki.unibe.ch/hermion/icons/>:

- What means "super method does not exit" ? Do you mean "exists"?

yep, this should read "does not exist", sorry.

- What has priority if there are multiple super sends in one method?

Priority has any super send to another selector than the one of the method. Afterwards we show whether the method sends super to the same selector, but there is no super method (means we don't check whether there is no implementor of a different super send selector as this gets highlighted in any case by the 'other selector' icon). Otherwise, we show the normal super send icon, distinguishing between a sub-impl. exists or not.

- What is shown when you have a sub-implementation and do a super call to a non existing method?

the icon that says "super send, but super method does not exist, w/ sub-impl." (triangle down, triangle up, red super sign)

Again there is an inconsistent mix of icons and meanings, for example:

- The case "super send, but super method does not exists." has a big yellow up arrow. The big yellow arrow on its own means "overrides super class methods", however this method does not exist in the given example.

ok, that's true, right. But this would imply that we have to do a completely different icon in this case which would not be consistent with the others. Hard to tell which inconsistency is worse.

For the test icons:

- The icons for "more red tests than green", "equal number of green and red tests" and "more green tests than red" are cute, but absolutely useless because inconsistant. "red", "yellow", "green" and "black" is the only information that is valuable. Or do you plan to add icons for all combinations like "some red tests, a few yellow and green and most of them are not run yet"?

We should probably take into account the not executed tests, I agree. Distinguishing between yellow and red would be nice too, but I think it's fair enough to just give feedback about the ratio between succeeded and not succeeded.

- Abstract test cases always remain in the state of not being run. Maybe it is better not to show the test icon at all in that case, or fetch the result from all the concrete subclasses (that would go nicely with the semantics of SUnit).

ok, good point. David