dbcli / odbc-cli Goto Github PK

Currently completion works well (or so I think!) in queries that are being typed out.

However there is one issue I have encountered - say:

• You connect to a database: DbMetadata object gets a list of catalogs as the connection is immediately expanded.
• You recall a query from history SELECT * FROM AdventureWorks2017.Person.Address
• Auto complete works fine if you were to, for example expand the query with ... WHERE.
• However if you were to erase from the end and try and reach a different table within the AdventureWorks2017.Person schema completion would come up blank since at that point it hasn't enumerated the schema within the AdventureWorks2017 catalog. (DbMetadata returns None, which is a signal to the completer that this is not a valid catalog.schema combination)

One possible solution - prior to returning None, perhaps DbMetadata should check if AdventureWorks2017 has been listed at all.

Long term solution is that all the extend_* calls should be taken out of the completer and brought inside DbMetadata so that we can house all the logic of the form if not found then query in one place.

sql_sidebar: myControl (buffer control) should only reconstruct the document if something in the sidebar has changed (expanded / collapsed). (my_app.obj_list has changed?)

sql_sidebar: myLexer, rather than storing the entire list of tokens, what about tokenizing only the specific line that was requested?

Hello,
I really liked this tools (mycli, ..etc).
It would be great to add support for Oracle DB.

Expand operations happen in a thread.

While in theory, these should be quick, need to insert some locks here to avoid races as we do surgeries on the object list (expand / collapse).