« Changelog History


PaperTrail v9.0.0 Release Notes

Release Date: 2018-03-26 // about 6 years ago

  • 💥 Breaking Changes, Major

    • #1063 - touch will now create a version. This can be configured with the :on option. See documentation section 2.a. "Choosing Lifecycle Events To Monitor".
    • ⬇️ Drop support for ruby 2.2, whose EoL is the end of March, 2018
    • PaperTrail now uses frozen_string_literal, so you should assume that all strings it returns are frozen.
    • Using where_object_changes to read YAML from a text column will now raise error, was deprecated in 8.1.0.

    💥 Breaking Changes, Minor

    • Removed deprecated Version#originator, use #paper_trail_originator
    • Using paper_trail.on_destroy(:after) with ActiveRecord's belongs_to_required_by_default will produce an error instead of a warning.
    • Removed the warn_about_not_setting_whodunnit controller method. This will only be a problem for you if you are skipping it, eg. skip_after_action :warn_about_not_setting_whodunnit, which few people did.

    🗄 Deprecated

    • #1063 - paper_trail.touch_with_version is deprecated in favor of touch.
    • #1033 - Request variables are now set using eg. PaperTrail.request.whodunnit= and the old way, PaperTrail.whodunnit= is deprecated.

    ➕ Added

    • #1067 - Add support to Rails 5.2.
    • #1033 - Set request variables temporarily using a block, eg. PaperTrail.request(whodunnit: 'Jared') do .. end
    • #1037 Add paper_trail.update_columns
    • #961 - Instead of crashing when misconfigured Custom Version Classes are used, an error will be raised earlier, with a much more helpful message.
    • Failing to set PaperTrail.config.track_associations will no longer produce a warning. The default (false) will remain the same.

    🛠 Fixed

    • #1051 - touch_with_version should always create a version, regardles of the :only option
    • #1047 - A rare issue where touch_with_version saved less data than expected, but only when the update callback was not installed, eg. has_paper_trail(on: [])
    • #1042 - A rare issue with load order when using PT outside of rails
    • #594 - Improved the error message for a very rare issue in the experimental association tracking feature involving two has_one associations, referencing STI models with the same base class, and the same foreign_key.