Beman Dawes wrote
The big question in my mind is whether we should enhance the current regression testing system or start fresh using one of the continuous integration frameworks?
I think we should think bigger. I'd like to see us encourage each user to run tests on the libraries he actually uses and post these results to a common area. This would mean that a) we would have test results for (all?) the actual configurations that users are using. b) The testing load be automatically be distributed to the users which actually use a library c) Libraries actually being used would be tested. d) Resources wouldn't be expended on libraries not being used. In short, distributing testing this way would scale to any number of libraries. I implemented a first cut of such a system as part of the Boost Incubator. This system uses CMake/CDash as described here: http://rrsd.com/blincubator.com/tools_cmak/ It permits users of libraries in the incubator to check out test results of other users. You can see how this works by going to http://rrsd.com/blincubator.com/bi_library/safe-numerics/?gform_post_id=426 and clicking the field in "Test Results Dashboard" In my view, a) Boost has to continue to grow or die. b) To continue to grow, it has to be scalable c) the best way to do that is to distribute testing to the users I see this a one of the ultimate consequences of the Boost modularization effort. Robert Ramey -- View this message in context: http://boost.2283326.n4.nabble.com/Enhancing-regression-tests-summary-tp4668... Sent from the Boost - Dev mailing list archive at Nabble.com.