Today, etcdctl endpoint health only performs a quorum GET. If server is corrupted or exceeded quota (which put the server into read only mode), this command incorrectly returns healthy.

While I try to change header titles by passing array of titles to options like below it does not override the headers. Instead it writes new headers first and original data with old headers again from next cell.

I am experiencing the same problem.

In addition, sheets js is mutating the header array passed in, which is not something I would ever expect.

Example:

const head

Requesting to add support for \connect database similar to psql format, to be used in addition to how USE database is currently used in CRDB.

Sometimes it is needed to store compressed data in the DB. Unfortunately not all the DBs have built-in compression and FUSE compressed FSes are not available for every OS. So it may make sense to store compressed binary blobs in the DB.

Unfortunately when one sees them in DBeaver he sees them compressed, but often they are needed uncompressed. So it'd be nice to have a feature to decompress the

Issue description

Go 1.15 will introduce a new Validator interface, which may be implemented by Conn to allow drivers to signal if a connection is valid or if it should be discarded:

https://tip.golang.org/pkg/database/sql/driver/#Validator

Today, we translate this as @value IS NULL OR LTRIM(RTRIM(@value)) = N''. But I have a few questions:

1. Why do we trim both the start and the end? Just one would produce the same result with less processing.
2. More importantly, why do we trim at all? As we know from our StartsWith research, N' ' = N'' returns true in SQL Server.

IMHO, the translation should be `@value IS NULL OR @valu

Like what was done in #943, it would be good if searchable attributes were also sorted in lexicographical order.