BUG:392112
FIXED-IN:3.2.0
Note: functionality of migration is not affected, it's just the code cleanup thanks to the new API.
BUG:392112
FIXED-IN:3.2.0
Note: functionality of migration is not affected, it's just the code cleanup thanks to the new API.
Build KEXI with this KDb: https://phabricator.kde.org/D11547
Open kdb.git/autotests/data/missingTableTest.kexi, it should not list persons table anymore.
Automatic diff as part of commit; lint not applicable. |
Automatic diff as part of commit; unit tests not applicable. |
Should we clean up the internal schema of non existing tables? Options would be automatic or prompt to user. Other consideration is other objects referring to non-existing tables like reports and queries.
I think I've been thinking about that. Well, mid-term we will only maintain internal metadata for backward compatibility. Since 3.2 by just ignoring the "ghost" internal metadata items we can ignore their existence, removing them won't change anything. Removing may be impossible (read-only mode can be enabled for example).
Other consideration is other objects referring to non-existing tables like reports and queries.
Yes, this will be always the case even after we no longer require creation of internal metadata. I think the answer at GUI level in KEXI would be feature that shows global relationships between all types of objects and help track and fix errors. Errors can appear less or more likely; more in a multi-developer project. Just like in text-based programming :)