Hi,
I like the work you have done, a lot. But there are a couple of things preventing me from using it more. Part of the library that I use as database layer needs support for:
Alter column: Rename a column, drop a column. I know that I can create a temp table to hold the data and create the new table with the new structure, but it is extremely inefficient. Renaming a column should be much more easily done by changing the name of the column in the internal structures, instead of moving data around. Dropping a column should be just a mark, even if the data is still in the file. The data can be eliminated at a later stage during a compact operation invoked by the user.
Updateable views: it would be very helpful if tables can be partition horizontally, then with a view, link the sister tables. The same way that SQL Server does with partition tables. Even better if the different tables can exist in different files, so when distributing the app, only the appropriate data for the user is distributed.
Also, is there any way to see the query plan?
I know that these are relatively big changes, but if they are possible to consider I would really appreciate it. And please let me know if I can help with some tasks.
Jorge Varas