Hello Stefano, On Wed, Jun 10, 2009 at 15:19, Stefano Zacchiroli<[email protected]> wrote: > Then, IMO, you should go back to step 1 of my answer. Given that you > are going to need per-package work anyhow, it would be better to push > that work upstream, providing per-package patches that do that.
Ok, that makes sense. But I'm wondering if such test-per-package would not clutter the package with unneeded (except for tests) binaries and/or files. On the other side, one could use examples and other provided files. I suppose some experiments would say if this is the case or not. The other side of the coin is to have a kind of standard to call all the tests of a package and interpret the results. Something like /usr/bin/<program>-run-all-tests. If I want to monitor the 246 OCaml packages in Debian/Ubuntu, I don't want to have 246 different ways to call the tests and get the results. Once again, some field experiment will tell what kind of requirements is needed. Many thanks to all of you for your inputs. Yours, d. -- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected]
