Comment 6 for bug 120392

Responding to Sebastien, yes, you have to delete some data before you see the problem. However, good design would not then make it impossible to delete the associated handle on the data. On the contrary, if the program notices something is missing it should either delete the handle to the missing data or (better) complain and offer options. Judging by the number of similar bugs, this is a common problem.

The idea that the ap directory is untouchable is a holdover from Windows and is quite non-Unix. There are lots of reasons to allow the user to work with data directly, and the better aps both document how to do this and handle the changes gracefully. In this case, one is *forced* to look under the hood to do the basic task of moving data from one machine to another. This basic task is totally undocumented, and the obvious action of simply moving a .evolution tree from one machine to another makes all your non-default calendars inaccessible from within the program, even though the data are still in the same files as on the old machine. You can only say "do nothing outside the program" if it is both possible and well documented how to do all the obvious tasks from inside the program. This is not currently the case: you do have to poke around in .evolution to find the files to import to regain access to the data.

--jh--