On 3/14/06, Adam Kennedy <[EMAIL PROTECTED]> wrote: > > > Even better would be adding two more states so that you can > > distinguish between prereq-failure, build-failure, test-failure and > > ok. > > Well actually I tried to sit down the other day and work out how many > distinct types of success/failure events there might be, for use in PITA. > > The number came out to something like 15 :) > > Which I imagine means we end up with something like a HTTP numbering system. > > And then for any given reporting system, depending on what you consider > "good", "bad" or "otherwise" in that particular system, you fold the 15 > result codes into your case-specific results.
That sounds all fine and useful, but a broad summary is also useful, just not when its over broad. Id say anything in the 3-5 range is ok, after that it gets to be too much unless you specifically are doing fine analysis. And i also agree with yuval that using numbers for this stuff is not so useful. Although i guess having both wouldnt be that horrible. Yves -- perl -Mre=debug -e "/just|another|perl|hacker/"