Mike Kupfer <mike.kupfer at sun.com> writes: > The tooltest changes look okay. It'd be nice to have a couple > additional test cases where there's no whitespace between the junk text > and the valid text. > > And I suppose you could add a test case with a multiline string if you > like. (This would be for protection in case of a bug somewhere else in > the system.)
Updated webrevs, the former is complete, the latter is relative to the originally posted webrev (~richlowe/6740590/) http://cr.opensolaris.org/~richlowe/6740590-2 http://cr.opensolaris.org/~richlowe/6740590-1_2 I think these deal with the points Mike raised, with a couple exceptions: The ARC regex doesn't really need change, we check the ARC synopsis, so a mis-spelled ARC name will show as not present in the DB, and be flagged anyway. Leaving things looser just leaves us needing to change code if ARCs are added or removed. The regex changes Mike wanted (\A and \Z), are only really applicable if we allow for multi-line comments (and set the MULTILINE flag on the expressions), we do not allow multi-line comments, and in fact do not wish to allow for them. Mike requested that, instead, we make this more explicit and throw an exception if we encounter a \n, which these changes do. I talked this out with dmarker, who gets comments in the same way that Cadmium would, and feels this would not be a problem for him. tooltest passes, cadtest passes. I'm still somewhat nervous about the possibility of problems with the gate's use of comchk, but I don't have anything concrete to base those concerns on. If someone were to find a way to test this code against the gate hooks, that would be great (or, Dave, if you could email me with how to configure your hooks such that they'd probably if I ran them locally, it's possible I could test it, I think?) -- Rich
