On 26/03/2008, Andy Armstrong <[EMAIL PROTECTED]> wrote: > On 25 Mar 2008, at 18:31, Ovid wrote: > > Is there any chance we can produce something viable in Oslo that's > > likely to solve this problem or is this a low enough priority in most > > people's minds that they're not bothered? I'm keenly focused on > > managing large-scale test suites and need stuff like this. > > > > I'm +1 on that. It's not too hard to add nested blocks to the parser - > in fact I've already done that experimentally. I got stuck on the API > design around how we represent nested results without overcomplicating > the existing parser interface. That would be a great thing to work on.
If you do this, there are lots of choices. Please make sure, that it is at least compatible with - each block having a plan - each block being counted as a single test in the plan of the outer block, the reasoning being that test_foo(bar, fuz) can change from being something which outputs a single test result to something which outputs a block of test results with no alterations to the plan containing the call to test_foo(). It also seems reasonable that the success value of a block == (success of all tests) AND (success of block plan), F > > -- > Andy Armstrong, Hexten > > > > >
