v2.1.6

• #87 Fix an issue where G/P/GP variants weren't generated for SetOps
• #88 Document the fact that the boil.Begin() feature exists better

v2.1.5

Bug fixes

• #77 Unique definitions were not correct in that if a unique key was placed over two columns, each column became unique which shouldn't be the case. This resulted in relationships generating one-to-one where they should have been to-many.
• #85 Fixed an oversight in Postgres where information_schema doesn't provide enough information to uniquely identify a duplicated constraint name causing multiple relationships to be generated.
• #84 Fixed a postgres-ism where no data is returned from a query when there was no data updated in an upsert which would return a spurious error.

v2.1.4

Add --tinyint-as-bool flag for MySQL. This flag tells the generator that you wish tinyint(1) columns to be mapped to Go's bool type opposed to the default type of int8.

v2.1.3

Fix missing StringArray import for generated structs that point to database arrays.

v2.1.1

• Fix an issue where combining query mods with relationships would break
• Fix an issue where during eager loading if you had: Load("One.Two") and One failed to return anything because there were no records, there would be a panic

v2.1.0

Features

• Add support for Enums in MySQL and Postgres
• Add support for single bytes ("char" type) for Postgres
• Upgrade to github.com/nullbio/null.v6

Bug Fixes

• Add the entire list of golint checked acronyms to sqlboiler's TitleCase (ID, UUID, etc.)
• Fix an issue where --basedir was being ignored in environment & command line (thanks pritambaral)
• Fix issue where eager loading N separate entities on a model would overwrite the previous N-1 loads

Upgrade Instructions

Please ensure you upgrade the null package to the latest version or you may run into trouble with this release.

go get -u gopkg.in/nullbio/null.v6

v2.0.5

• Add "ip" to uppercase words for struct members
• Add new line to warnings for incompatible data-type to improve readability

v2.0.4

• Fix an issue where MySQL views were being generated as tables

v2.0.3

This release addresses issues with fairly specific use cases, but important ones. Many thanks to jseriff for finding most of these bugs.

• Added guaranteed .R struct presence while eager loading to reduce (ironic) nil-check boilerplate while eager loading.
• Added a --version flag
• Fix an issue in Postgres where constraints in other schemas affected the current one which would generate extraneous relationships
• Fix an issue where eager loading would not properly load nested models
• Fix an issue where eager loading would be overridden by subsequent calls to qm.Load()

v2.0.2

Overall these changes make for more robust naming and generation of tables with less seen relationships. An example is a self-referential table, or a join table that has extra columns + joins to itself. These situations caused some generated SQL to fail as well as some problems with name collisions in generated code.

• Correct some cases of mysql insert that would fail to generate correct sql
• Change the naming of certain relationships to preserve uniqueness
  • This may have broke some function names in existing code
• Rather than gofmt individual template generation, run it on the entire output file to remove excess newlines
• Stop postgres tests from prompting for passwords that are in the sqlboiler config
• Fix generation of many-to-many with self-referential tables
• Fix SetOp for join tables that have more than 2 columns
• Fix non-join tables being flagged as such which in turn didn't generate tables for them
• Refactor output section of code to be more efficient
• Update CI schemas to include more cases for relationships