about summary refs log tree commit diff
path: root/config/initializers/0_duplicate_migrations.rb
diff options
context:
space:
mode:
authorThibaut Girka <thib@sitedethib.com>2019-11-20 17:18:00 +0100
committerThibaut Girka <thib@sitedethib.com>2019-11-20 17:18:00 +0100
commit33c2a7e23c3e53f05ac93ec533e2c9b238e2cc34 (patch)
tree9dd895aaffa9759964327c774eb7ea67a7366897 /config/initializers/0_duplicate_migrations.rb
parent6475525f3176698d6d8902b32af62ffd23ddd9e3 (diff)
Add documentation about the migration hack
Diffstat (limited to 'config/initializers/0_duplicate_migrations.rb')
-rw-r--r--config/initializers/0_duplicate_migrations.rb14
1 files changed, 14 insertions, 0 deletions
diff --git a/config/initializers/0_duplicate_migrations.rb b/config/initializers/0_duplicate_migrations.rb
index 4c7440fa4..d15d2b24a 100644
--- a/config/initializers/0_duplicate_migrations.rb
+++ b/config/initializers/0_duplicate_migrations.rb
@@ -1,3 +1,17 @@
+# Some migrations have been present in glitch-soc for a long time and have then
+# been merged in upstream Mastodon, under a different version number.
+#
+# This puts us in an uneasy situation in which if we remove upstream's
+# migration file, people migrating from upstream will end up having a conflict
+# with their already-ran migration.
+#
+# On the other hand, if we keep upstream's migration and remove our own,
+# any current glitch-soc user will have a conflict during migration.
+#
+# For lack of a better solution, as those migrations are indeed identical,
+# we decided monkey-patching Rails' Migrator to completely ignore the duplicate,
+# keeping only the one that has run, or an arbitrary one.
+
 ALLOWED_DUPLICATES = [20180410220657, 20180831171112].freeze
 
 module ActiveRecord