DBeaver Forum

Hi,

I just discovered dbeaver and love it, apart from one big bug: the autocomplete does not work for me.

I am connectiong to Vertica, a columnar DB, via a JDBC driver, provided by them.

I can connect, execute statements, look at the schema tree without problem.
I can word-complete (shift+trl+space), no problem.

Unfortunately, auto complete in a script from a table name or alias does not work (even when I am properly connected). I can see that dbeaver tries to do something (wait icon appears) but nothing happens, and I cannot se anything in the logs.

I would love to see this fixed, and am ready to provide any help at my end, but I am not sure wht more to add.

Thanks,
Guillaume

I just noticed that if I use a full name for a table, for instance:
dbname.schemaname.dbname I can then have completion on each part of the name.

To be sure, I even set up the advanced parameter of the driver (get active database), but it did not change a thing.

Would that be a bug or a feature?

Thanks,

Hi!

Generally autocomplete is a pretty tricky thing. There are several different cases: (_ is a place where you press ctrl+space)
- SELECT * FROM _ (will show all tables in current schema and all schemas in current database)
- SELECT * FROM SCHEMA._ (all tables in schema SCHEMA)
- SELECT * FROM abc_ (all accessible tables with name like "abc%")
- SELECT * FROM SCHEMA.abc_ (tables in schema SCHEMA with name like "abc%")
- SELECT a._ FROM TABLE a (columns of table TABLE)
Also any identifier might be quoted (e.g. "TableName") - in this case DBeaver will use case sensitive objects search.

I never worked with Vertica database but as I understood it has unusually complex table name structure: DATABASE.SCHEMA.TABLE
Most databases have only one hierarchy level - schema or database.
Probably there is a bug in DBeaver in case of such complex table names. I'll investigate this issue.

Thanks for report.

Hi Serge,

Thanks for your answer.

Indeed Vertica can use the database.schema.table syntax (just like postgres actually), but this is for compliance with standard SQL only, and the 'database.' bit is not needed.
Vertica is pretty close to postgres (they both started as projects from the same researcher), maybe this can help?

Thanks,

Umm. If database part is not required (in Postgres actually there is only one database accessible with one physicall connection so db name is never used in queries) then evrything should work ok. It seems that the problem is in JDBC driver.
In fact DBeaver contains many small workarounds for different databases in Generic plugin (as this plugin supposed to work with any JDBC-compliant database). And unfortunatelly driver vendors often implement JDBC API on their own specific way.
And there are a few things specific for PostgreSQL. Maybe we need some more for Vertica.

I'll try to test autocomplete on CE or evaluated version of Vertica. And, if possible, next version of DBeaver will contain a fix of this issue.

Fantastic. If I can help do not hesitate to ask.

As for PG, there can be only one DB accessed by connection, so the database name is never actually used either.

Thanks,
Guillaume

Hi Serge,

Did you have any chance to look at this, or is there anything I can do to assist?

Thanks,

Hello,

Unfortunatelly currently I've no access to working instance of Vertica. Mostly all our testing environment hosts on windows. I've tried to set it up on Ubuntu vm but something went wrong..
If there is any chance to gain access to any test/dummy Vertica server - I'd be very appreciated. If it is possible - please write me on serge@jkiss.org.

But probably we'll have Linux based server testing environment till the end of year and I'll try to set it up one more time.

Thanks.

Hi Serge,

I talked with the power that be at $work and they are happy to set up a testnig Vertica instance for you to try out.
There is currently some issues which unfortunately have higher priorities, but I'll get back to you with connections details as soon as possible.

Thanks,
Guillaume

Hi Guillaume,

That'd be great!