On Mon, Mar 03, 2003 at 10:32:04AM -0500, Sam Ruby wrote: > Stefano Mazzocchi wrote: > > > >Yes, preaching to the converted. I think that Forrestbot can be the only > >serious unit test that can be useful to Cocoon. > > I am? > > <joke>Is there a word in Italian for people who assert that they are > pious, but you don't tend to see them in church all that often?</joke> > > Forrestbot is a specific example. When that effort started, I tried to > see if there was an potential synergy with Gump, but the perception I > got was that that effort was exclusively about publishing and not about > regression testing.
That is mostly true, although I've subsequently written add-on shell scripts to always use the latest version of Forrest. It would be fairly easy to throw in latest versions of the Cocoon and Avalon jars, fetched from wherever Gump puts them. I don't think Forrest would make a very good Cocoon test suite though, because: - Forrest doesn't exercise even 5% of what Cocoon offers. - Forrest is trying to provide a service to users, not keep up with bleeding edge Cocoon. I think that resurrecting the Anteater[1] test scripts is a much better plan. Each block could have a tests/ directory, with a handful of Anteater scripts exercising the samples in that block. I once had a fairly complete Cocoon test suite running (and 80% passing), but the samples layout kept changing and it gradually became out of date. --Jeff [1] http://aft.sf.net > My current "preaching" is an attempt to restart that discussion. > Stefano, given your recent interest in Gump, now seems like an opportune > time. > > I don't care if it is run on my machine or on somebody else's. I don't > care if any gump code is used in the process. But I do think it would > be helpful to run Forrest with up to the minute versions of everything > in an effort to get more people talking, and talking earlier, with > developers of components that they depend on. > > And I am willing to help. > > - Sam Ruby > >
