0.7a0

• Fixed bug where the item table did not correctly link to the commits using a foreign key. #59
• Fixed bug where some repositories would not process correctly due to data from older versions not being successfully loaded from the commit history. #64

0.6.1

• Fixed bug where databases containing multiple file histories created using the --namespace option could have later commits applied to them in subsequent runs. #42, #43

0.6

• Fixed critical bug where columns were incorrectly recorded as consistently toggling between null and their current value. #33
• Documentation now includes links to live examples of databases created using this tool. #30
• --wal option for turning on SQLite WAL mode - useful if you want to safely run queries against the database file while it is still being built. #31
• Fixed bug where list and dict values were not correctly compared for equality. #32
• The item_version_detail SQL view now includes a _changed_column JSON array of column names that changed in each version. #37
• Nested packages such as --import xml.etree.ElementTree can now be imported. #39
• item_version._item is now an indexed column. #38

0.5

• The item_version table now only records values that have changed since the previous item version. A new item_changed many-to-many table records exactly which columns were changed in which item version, to compensate for ambiguous null values. #21
• New --full-versions option for storing full copies of each version instead of storing just the columns that have changed.
• Major backwards-incompatible schema changes - see README for details of the new schema.
• New --dialect option for specifying a CSV dialect if you don't want to use auto-detection. #27
• The history for multiple files can now be stored in the same database, using the new --namespace option. #13
• --skip HASH, --start-at HASH and --start-after HASH options for skipping specific Git commits or starting processing at or after a specific hash. #26, #28

0.4

• Major changes to the database schema. Foreign keys now use integer primary key IDs rather than using lengthy item or commit hashes, which reduces the database size for large repositories by almost half. #12
• Python generators can now be used in --convert functions. #16
• Reserved columns are now marked by an underscore prefix, for example _id and _commit. #14

0.3.1

• Fixed bug where files with a "rowid" column would fail to import correctly. #10

0.3

• git-history file command now shows a progress bar, unless run with the --silent option. #9

0.2.1

• Improved usage documentation in the README

0.2

• --csv option can now be used to process CSV or TSV data - the dialect is detected automatically. #6
• Implemented reserved column names: id, item, version and commit. If your data includes any of these column names they will be renamed to id_, item_, version_ and commit_ respectively. #8

0.1

• Initial release. git-history file db.db filepath.json command, see README for details.