Re: [Jikesrvm-core] [Jikesrvm-regression] FAILED 36 tests

[Steve B. <Ste...@an...>]

On 17/05/2007, at 9:46 PM, Peter Donald wrote:
> ug. I butchered the top of 2 files when I added the license header.
> Thus the tests didn't compile and thus they were not reported in xml.
> Should be fixed now.
No problem. Thanks.

>> These tests don't show up as failures because they don't appear at  
>> all.
>
> maybe we present absent tests in different color and list them
> separately, same for new tests?

In the table (which seems to be what caught Dave's eye), I could  
color in orange any tests for which none were run, but for which 1 or  
more were run on other days that week.  I'll take a look at doing  
this while I check what's going on with the "new failure" line. I'm  
not 100% sure there is actually a problem, and it's hard to see how  
that part of the script could have rotted, but I'll try to find a  
concrete example of it going wrong and start from there.

Putting a high level "Did not run N tests" summary is harder since  
right now the xml does not tell me (AFAIK) what tests were not run.   
I only know what passed and what failed.  I can only infer missing  
runs comparing with other days.

--Steve