Le 29/01/16 07:41, Zheng, Kai a écrit : > Thanks for the great work, Emmanuel. > > I have some questions. As discussed previously, it's hard to recreate the > database corrupt issue. Did you include a test with a sample corrupt database > image and show if it's can be repaired? There is no need of a corrupted database to test it. The assumption is that the MasterTable is not corrupted (because if it is, we are dead in the water...). Based on this assumption, we simply delete all the existing indexes, and recreate them from scratch.
Testing it is just a matter of creating a database, run the repair tool on it, and check that all the resulting content is correct. This is what the unit test is doing.
