Releases · rroblak/seed_dump

What's New

This is a major maintenance release that addresses many long-standing issues and adds new features.

UPSERT_ALL option (#104): Preserves original record IDs using Rails 6+ upsert_all, fixing foreign key references becoming invalid after reimport
INSERT_ALL option (#153): Bulk insert using Rails 6+ insert_all for faster imports
INCLUDE_ALL option (#147): Include all columns, cleaner alternative to EXCLUDE=""
MODEL_LIMITS option (#142): Per-model limit overrides (e.g., MODEL_LIMITS="Teacher:0,Student:50")
HEADER option (#126): Add a comment header with timestamp and copyable rake command to seed files

Foreign key ordering (#78, #83): Models now dumped in dependency order based on belongs_to associations
STI deduplication (#120): STI subclasses no longer create duplicate seed entries
HABTM deduplication (#26, #114): HABTM models sharing the same join table are deduplicated
HABTM private constants (#130): Use const_get for HABTM join models to avoid NameError on import
Timezone preservation (#111): Use ISO 8601 format for datetime dumps to preserve timezone info
ActionText support (#154): Handle ActionText::Content in seed dumps correctly
Tables without primary key (#167): Handle ordering for tables without a primary key (also fixes #138, #123 duplicate records bug)
default_scope with select (#165): Handle default_scope with select in COUNT queries
Nested Time/BigDecimal (#105): Handle Time/BigDecimal values in nested Hashes and Arrays
Non-model AR descendants (#112): Handle abstract classes and other non-model ActiveRecord::Base descendants
Model names ending in 's' (#121): Fix incorrect singularization (e.g., "Boss" → "Bos")
Pipe compatibility (#150): Use write-only file modes for /dev/stdout and pipe compatibility
CarrierWave uploaders (#117): Uploader columns now dump correctly instead of nil
Batch size output (#127): BATCH_SIZE now outputs separate method calls per batch

Full Changelog: v3.3.1...v3.4.0