I think people are using database_location and it has never been validated in any way until now. This is bad because we use its directory as cwd for subprocess.Popen calls. I think what was happening was database_location was either invalid or else we weren't expanding ~ to a full path. Either way, I think what was failing was the cwd argument to subprocess.Popen. The call would fail and it would look like the cscope path was wrong, but I think that was a red herring. Added lots of validation to database_location, so hopefully this helps if not fixes the problem. It should at least make it more clear whether that's the problem or not.
I think people are using database_location and it has never been validated in any way until now. This is bad because we use its directory as cwd for subprocess.Popen calls. I think what was happening was database_location was either invalid or else we weren't expanding
~to a full path. Either way, I think what was failing was the cwd argument to subprocess.Popen. The call would fail and it would look like the cscope path was wrong, but I think that was a red herring. Added lots of validation todatabase_location, so hopefully this helps if not fixes the problem. It should at least make it more clear whether that's the problem or not.