FeatureBase v3.35.0

About

This is FeatureBase v3.35.0, a crazy fast database/analytics engine based on Roaring Bitmaps.

FeatureBase is written in Go and supports SQL.

License

FeatureBase is released under the Apache 2.0 Open Source license.

Support

Join the Discord community: https://discord.com/invite/featurefirstai

Documentation: https://docs.featurebase.com/

Cookbooks (including SQL bulk inserts): https://github.com/FeatureBaseDB/featurebase-examples

Would you rather use Docker?

Release

This release contains AMD and ARM FeatureBase binaries for macOS (darwin) and Linux.

The tarball for each architecture includes a directory for FeatureBase and an IDK directory for consumers for CSV, Github, Kafka and SQL.

Highlights

Added support for NULL in non BSI fields, improved test coverage and included bug fixes. Future version numbers of the FeatureBase binary will be increased to match our internal bundle version numbers, so keep an eye out for this soon.

Resolved Bugs

• FB-2065 Selecting from View defined using CURRENT_TIMESTAMP in where clause will throw error
• FB-2090 No error for misconfigured fields

NOTE: The tickets/labels above and below are for our reference to Jira issues and do not correspond to issues on Github.

Task

• FB-1689 Add an _exists bitmap for non-BSI fields.
• FB-2015 Grafana: Ability to Use Native Time Filter
• FB-2093 CREATE/etc USER, GRANT/REVOKE
• FB-2027, FB-2036, FB-2039, FB-2040, FB-2041, FB-2045, FB-2046, FB-2049, FB-2054, FB-2055, FB-2056 and FB-2074 Add SQL3 tests

Issues

• Dashboard UI still isn't showing metrics on imports
• Delete can delete records it probably shouldn't

Comments

Keep in mind we don't support all SQL statements, yet.

FeatureBase v3.34.0

About

This is FeatureBase v3.34.0, a crazy fast database/analytics engine based on Roaring Bitmaps.

FeatureBase is written in Go and supports SQL.

License

FeatureBase is released under the Apache 2.0 Open Source license.

Support

Join the Discord community: https://discord.com/invite/featurefirstai

Documentation: https://docs.featurebase.com/

Cookbooks (including SQL bulk inserts): https://github.com/FeatureBaseDB/featurebase-examples

Would you rather use Docker?

Release

This release contains AMD and ARM FeatureBase binaries for macOS (darwin) and Linux.

The tarball for each architecture includes a directory for FeatureBase and an IDK directory for consumers for CSV, Github, Kafka and SQL.

Highlights

This release continues to build on support and GA release for SQL functionality. Future version numbers of the FeatureBase binary will be increased to match our internal bundle version numbers. A major version bump is needed due to the GA release of SQL.

Resolved Bugs

• FB-1915 Large ID type data inserted with incorrect values
• FB-1916 Large INT type data inserted with incorrect values
• FB-2057 Gracefully handle divide by zero in SQL
• FB-2062 ORDER BY failing with larger data sets
• FB-2061 Internal error triggered by SQL with grouping and Complex Conditions
• SUP-308 Avg() returns incorrect result when selection returns zero rows

NOTE: The tickets/labels above and below are for our reference to Jira issues and do not correspond to issues on Github.

Task

• FB-1843 / FB-1846 Kafka Avro Delete Consumer Functionality (delete full records, delete by value)
• FB-1893 Date/Time scalar function - datetimeadd
• FB-1894 Date/Time scalar function - datetimediff
• FB-1895 Date/Time scalar function - datetimefromparts
• FB-1896 Date/Time scalar function - datetimename
• FB-1897 Date/Time scalar function - datetimepart
• FB-1961 Date/Time scalar function - date_trunc
• FB-1952 Kakfa Bulk Insert Tooling

Issues

• Dashboard UI still isn't showing metrics on imports
• Delete can delete records it probably shouldn't

Comments

Keep in mind we don't support all SQL statements, yet.

FeatureBase v3.33.0

About

This is FeatureBase v3.33.0, a crazy fast database/analytics engine based on Roaring Bitmaps.

FeatureBase is written in Go and supports SQL.

License

FeatureBase is released under the Apache 2.0 Open Source license.

Support

Join the Discord community: https://discord.com/invite/featurefirstai

Documentation: https://docs.featurebase.com/

Cookbooks (including SQL bulk inserts): https://github.com/FeatureBaseDB/featurebase-examples

Would you rather use Docker?

Video

Video is forthcoming.

Release

This release contains AMD and ARM FeatureBase binaries for macOS (darwin) and Linux.

The tarball for each architecture includes a directory for FeatureBase and an IDK directory for consumers for CSV, Github, Kafka and SQL.

Highlights

The SQL3 feature flag has been removed, and the SQL3 is now the default. We've added new functions such as time quantum and aggregate functions.

NOTE: We use "SQL3" to refer to the new SQL support for FeatureBase. "SQL3" does not refer to a particular flavor of SQL.

Resolved Bugs

• FB-1887 fb_exec_requests query group by not working
• FB-1909 Bug in backup and backup_tar field key translation
• FB-1912 Restore uses an obnoxious amount of memory
• FB-1954 Weird capitalization column misrepresentation bug (Nope, not capitalization)
• FB-1963 Featurebase server error printed despite successful query
• FB-1968 Timestamp in bulk import not handling seconds properly
• FB-1974 SQL3: Select count fails when run against table with no _id
• FB-1980 Test timeout observed in CI
• FB-1997 HAVING producing incorrect results for some queries
• FB-2002 HAVING sum(decimal) gets type coercion error, possibly sum(decimal) is coming back as int64
• FB-1790 Resolved a Protobuf Error which prevented connecting Grafana to FeatureBase

NOTE: The tickets/labels above and below are for our reference to Jira issues and do not correspond to issues on Github.

Task

• FB-1558 Implement semantics to support inserts into time quantum fields
• FB-1654 Implement select on a timequantum nested table
• FB-1658 Factor out AggregateFunctionType
• FB-1852 SQL3 w/ Grafana Plugin
• FB-1955 Support aggregates on top of results/expression of other functions (i.e. AVG(len(col)))
• FB-1975 Clean up version check code. Add options for where to store the UUID and where to check for version.
• FB-1996 Limit memory allowed while creating backup
• FB-2013 Remove SQL3 feature flag

Active Bugs

None

Issues

• Dashboard UI still isn't showing metrics on imports (previously a bug, but we're kicking around getting rid of the current dashboard, so we moved it here)
• Delete can delete records it probably shouldn't (this is limited to pending inserts and is a hard problem with filtering we're advertising here while we investigate)

Comments

The previously required flag for enabling full SQL support is no longer required and this message will be removed on the next release.

Keep in mind we don't support all SQL statements, yet.

FeatureBase v3.32.0

About

This is FeatureBase v3.32.0, a crazy fast database/analytics engine based on Roaring Bitmaps.

FeatureBase is written in Go and supports SQL.

License

FeatureBase is released under the Apache 2.0 Open Source license.

Support

Join the Discord community: https://discord.com/invite/featurefirstai

Documentation: https://docs.featurebase.com/

Cookbooks (including SQL bulk inserts): https://github.com/FeatureBaseDB/featurebase-examples

Would you rather use Docker?

Video

Upon further review, this release was a short sprint, so we'll save our video editing skills for the next release!

Release

This release contains AMD and ARM FeatureBase binaries for macOS (darwin) and Linux.

The tarball for each architecture includes a directory for FeatureBase and an IDK directory for consumers for CSV, Github, Kafka and SQL.

Highlights

This release further enhances our SQL support by adding FORMAT, STR and other SQL string functions. In addition, release includes bug fixes for various functions.

Resolved Bugs (previously labeled Fixes)

• FB-1790 Resolved a Protobuf Error which prevented connecting Grafana to FeatureBase
• FB-1811 SQL3 CLI - NDJSON, BULK INSERT parsing errors
• FB-1903 fb_performance_counters cannot be queried on multi-node databases
• FB-1935 SQL AVG() Returns Incorrect Results
• FB-1938 SHOW COLUMNS internal data type and data type matching
• FB-1939 SQL3 > BETWEEN statement errors
• FB-1940 SELECT... WHERE returns same data as
• FB-1951 INSERT INTO with timestamp does not respect temporal reality
• FB-1953 Support MIN/MAX for STRING
• FB-1964 fb_cluster_nodes "space_used" column is a bool
• FB-1966 Flatten SQL object name casing when processing DDL statements.
• FB-1969 Uncaught Nil pointer error triggered for bulk insert
• FB-1970 The Null data in the CSV file caused the whole bulk insert fail with error
• FB-1978 "query error: table exceeds available memory" when running select count(*) from equipmentbig

NOTE: The tickets/labels above and below are for our reference to Jira issues and do not correspond to issues on Github.

Task (previously labeled Features)

• FB-1817 Implement FORMAT() function for SQL3
• FB-1862 Implement STR() function for SQL3
• FB-1899 Add ability to 'handle' bulk insert where server does not direct access to source
• FB-1907 Docs Review - SQL Consumer Overhaul

Active Bugs

• FB-1933 Bug in backup and backup_tar field key translation
• FB-1968 Timestamp in bulk import not handling seconds properly

Issues

• Pending SQL GA, requires using flags to enable new SQL support (--sql.endpoint-enabled)
• Dashboard UI still isn't showing metrics on imports (previously a bug, but we're kicking around getting rid of the current dashboard, so we moved it here)
• Delete can delete records it probably shouldn't (this is limited to pending inserts and is a hard problem with filtering we're advertising here while we investigate)

Comments

To run the server with the new SQL preview support, please use the following command (the docker builds in featurebase-examples do this for you):

./featurebase server --sql.endpoint-enabled

Keep in mind we don't support all SQL statements, yet.

FeatureBase v3.30.0

About

This is FeatureBase v3.30.0, a crazy fast database/analytics engine based on Roaring Bitmaps.

FeatureBase is written in Go and supports SQL.

License

FeatureBase is released under the Apache 2.0 Open Source license.

Support

Join the Discord community: https://discord.com/invite/featurefirstai

Documentation: https://docs.featurebase.com/

Examples: https://github.com/FeatureBaseDB/featurebase-examples

Would you rather use Docker?

Video

This release marks the first and ongoing bi-weekly sprint reviews we'll be publishing to YouTube.

]

Release

This release contains AMD and ARM FeatureBase binaries for macOS (darwin) and Linux.

The tarball for each architecture includes the consumers for CSV, Github, Kafka and SQL.

Notes are provided for guidance but are not yet inclusive of all features, fixes, issues and bugs. We're working on rectifying this.

Features

• Implements REPLICATE() for SQL
• String value support for Arrow (dataframes)
• Implements string method CHARINDEX() for SQL

Fixes

• SQL MIN() and MAX() don't work for timestamps*
• < and > operators don't work for timestamps*
• memory issues on bulk insert*
• ! → Bit-wise NOT operator doesn't work as expected*
• Large INT type data inserted with incorrect values*
• Restore uses an obnoxious amount of memory*

*previously not reported as a bugs, but they are fixed so there's that. 🪂

Issues

• Pending SQL GA, requires using flags to enable new SQL support (--sql.endpoint-enabled)
• Dashboard UI still isn't showing metrics on imports (previously a bug, but we're kicking around getting rid of the current dashboard, so standby)
• Delete can delete records it probably shouldn't (this is limited to pending inserts and is a hard problem with filtering we're advertising here while we investigate)

Bugs

• SQL AVG() Returns Incorrect Results (for fields of type ID)
• Bug in backup and backup_tar field key translation (cluster mode ONLY)
• --offset-mode flag doesn't change behavior

Comments

Moving forward, all work to FeatureBase will occur on this repository. 🚀

To run the server with the new SQL preview support, please use the following command (the docker builds in featurebase-examples do this for you):

./featurebase server --sql.endpoint-enabled

Keep in mind we don't support all SQL statements, yet.

FeatureBase v3.27.0

About

This is FeatureBase v3.27.0, a crazy fast database/analytics engine based on Roaring Bitmaps. It is written in Go.

FeatureBase is now released under the 2.0 version of the Apache License.

Support

Join the Discord community: https://discord.com/invite/featurefirstai

Documentation: https://docs.featurebase.com/

Would you rather use Docker?

Release

This release contains AMD and ARM FeatureBase binaries for macOS (darwin) and Linux.

The tarball for each architecture includes the consumers for CSV, Github, Kafka and SQL.

Features

• Implements SUBSTRING(), SPACE(), REPLACEALL(), UPPER(), TRIM(), RTRIM() and LTRIM() for string manipulation
• Implements DISTINCT() - note this was previously supported in the older SQL implementation and then stopped working in later builds with --sql-endpoint-enabled set.
• Implements DELETE() 🚀

Fixes

• Fixed Community UI to point to old SQL backend if newer SQL endpoint is not available
• UI: select * query erases history
• SQL CLI - NDJSON, BULK INSERT parsing errors

Issues

• requires using flags to enable new SQL support (--sql.endpoint-enabled)

Bugs

• Delete can delete records it probably shouldn't.
• Dashboard UI isn't showing metrics on imports

Comments

This release should be the last release that uses cherry picking from our private repo. We are planning on moving all commits to the public repo moving forward.

This release's most exciting features are the reimplementation of DISTINCT and DELETE in the new SQL preview, which is accessed using the --sql.endpoint-enabled flag. This will likely be the last release that requires use of that flag for previewing SQL support.

Keep in mind we don't quite support all SQL statements, yet.

FeatureBase v3.26.0

About

This is FeatureBase v3.26.0, a crazy fast database based on Roaring Bitmaps. It is written in Go.

Release

This release contains AMD and ARM FeatureBase binaries for macOS (darwin) and Linux.

The tarball for each architecture includes the consumers for CSV, Github, Kafka and SQL.

Support

Join the Discord community: https://discord.com/invite/featurefirstai

Documentation: https://docs.featurebase.com/

Would you rather use Docker?

Features

• SQL support now in the UI

Fixes

• help contains two holder command references (this is a new command that starts and exits the server)
• DDL statements on tables with hyphens on their names fail with a tokenizer error. (do not use '-' in names)
• Syntax errors may be triggered by running TOP and DISTINCT
• Top(x) Does Not Work with Additional Expression Clauses (WHERE, GROUP BY, etc.)
• SQL - Multiple Group By Columns Can Cause DB Health Issues
• SQL support not in the UI
• errors with ingesting CSV files with headers
• Bulk JSON imports
• Top() filter with AVG() errors
• Top() filters with count() errors

Issues

• requires using flags to enable revamped SQL support (--sql.endpoint-enabled)

Bugs

• UI: select * query erases history
• "IS" SQL Argument returns incorrect error
• SQL3 CLI - NDJSON, BULK INSERT parsing errors

Version Comments

Previous community version numbers were unfortunately chosen, including v1.0.0-community, v1.3.0-community, etc. Moving forward, the community release will follow the binary release versions.

It's complicated, but FeatureBase was originally named Pilosa and was very much Open Source. We closed the source for a time to do enterprise releases under the Molecula brand, running through a whole bunch of different versions. After some careful thought, we've opened up the code base again and arrived at using FeatureBase for both the company and the repository.

This should allow the homebrew recipe to be reactivated.

SQL Comments

We previously referred to "SQL3" in the releases here, but that was the internal name of the project to add most of the common SQL commands to the product, not the narrow implementation of SQL we previously supported (which we called SQL1). As of today FeatureBase now supports most SQL commands in both the UI and API calls to the '/sql' endpoint.

Keep in mind we don't quite support all SQL statements, but when asked our VP of Engineering said, "There's not much left to do...but the work remaining includes common table expressions (CTEs), windowing functions, UPDATEs (use INSERT REPLACE instead) and DELETEs, some of which which will be dropping soon."

v1.3.0-community

About

This is FeatureBase, a crazy fast database based on Roaring Bitmaps. It is written in Go.

Release

This release contains AMD and ARM FeatureBase binaries for macOS (darwin) and Linux.

The tarball for each architecture includes the consumers for CSV, Github, Kafka and SQL.

NOTE: The UI does not support SQL3 in this version. Please use the APIs or CLI instead.

Support

Join the Discord community: https://discord.com/invite/featurefirstai

Documentation: https://docs.featurebase.com/

Would you rather use Docker?

Features

• SQL3 support
• inserts via SQL3

Fixes

• Grafana issues

Bugs

• help contains two holder command references (this is a new command that starts and exits the server)
• DDL statements on tables with hyphens on their names fail with a tokenizer error. (do not use '-' in names)
• Syntax errors may be triggered by running TOP and DISTINCT
• Top(x) Does Not Work with Additional Expression Clauses (WHERE, GROUP BY, etc.)
• SQL3 - Multiple Group By Columns Can Cause DB Health Issues
• "IS" SQL Argument returns incorrect error

Issues

• UI does not support SQL3 in this release (in progress)
• requires using flags to enable SQL3

v1.1.0-community

Highlights

This release contains new SQL language support that can be enabled
using the sql.endpoint-enabled=true configuration option. This
feature includes a new command line interface (featurebase cli), a
new API endpoint (POST /sql), and greatly expanded SQL language
support (SHOW TABLES/COLUMNS, CREATE/ALTER TABLE, SELECT, INSERT
etc.). Documentation on this feature can be found in the Reference
section at docs.featurebase.com. This release also removes support
for Postgres Wire Protocol. ** NOTE: This feature should be considered
a preview and it is not recommended for production use. **

Bug

• Grafana: Replace empty variables to All()
• Kafka consumer stops reading messages from topic.
• Timestamp fields not supporting range queries
• UI - fix reported security issues

Features

• Restore previous backup/restore via stdout functionality
• release of new SQL language support in preview.

v1.0.0-community

This is v1.0 of the community release of Featurebase. It is built from commit da9b57b.