On Sat, 23 Oct 2004, Andre Poenitz wrote: > I think we should try to make an effort to get 1.4.0cvs into a state > where we could ask adventurous users to try it out.
I think that sounds like exactly the right thing to do. Regarding the assertions: That sounds like a brillant idea. If possible, make it so that debugging info is dumped on the console, but often assertions are just as wrong as the code itself, so it makes sense to knock them out and hope for the best. Another idea would be to implement a monkey-tester: It clicks randomly all over the place many times a second. (Just be sure to disable print and save.) A more efficient alternative would be a monkey tester would be to run random lyx funcs a lot of times on a big document, and repeat until crash. Then fix crash, and repeat. I've used such an approach with great success when stabilising an unstable product. Normally, it takes forever to find the bugs by hand, reproduce them as a minimal test case, and all that. With a monkey tester, you can leave some of the work to the machine: Brute force will find the bugs eventually. Regards, Asger
