0.4.1: Add support for arbitrary SQL statements

Signed-off-by: Byron Ruth <b@devel.io>

Snapshots

This release supports the -snapshot option which produces a set of row-stored events that contain a snapshot of row data in a table. In an event sourcing context, this enables both initiating state from the source table, but also snapshotting over time to decrease the amount of time a new consumer requires in order to replay the events (by starting at the most recent snapshot and reading forward).

See here for more details.

New events API and support for column renaming

Add table-level column renaming feature (#8)

Signed-off-by: Aaron Browne <aaron0browne@gmail.com>

0.3.1: Drop SQLite dependency for simple sort implementation

It was assumed that the SQLite implementation would be faster due to
inherent optimizations, but this was completely false.

For example, a ~166 MB CSV took ~1m13s with the SQLite implementation
and now takes ~12s with pure Go solution. It also takes a bit ~10% less
memory.

The other advantage is C Go is no longer needed so Windows builds are
available again.

Signed-off-by: Byron Ruth <b@devel.io>

Add support for unsorted CSV files

Two new options are provided -csv1.sort and -csv2.sort, which if either
are set, will sort the corresponding data by the -key columns before
diffing with the other source.

This uses SQLite behind the scenes which requires CGO. This release drops
Windows binaries for the time being until an alternate implementation is
considered.

Signed-off-by: Byron Ruth <b@devel.io>

Add support for pre-sorted CSV tables

Add support pre-sorted CSV tables

Signed-off-by: Byron Ruth <b@devel.io>

Add deleted_rows to diff output

Add Dockerfile

Signed-off-by: Byron Ruth <b@devel.io>

Fix bytes encoding

Convert bytes to string for output

Signed-off-by: Byron Ruth <b@devel.io>

Initial release

Initial commit

Signed-off-by: Byron Ruth <b@devel.io>