When you use Ruby 2.7.0’s new beginless range feature in a hash argument to .where the results it does not match the behavior of using Float::INFINITY..n.
irb(main):030:0> Post.where(id: Float::INFINITY…3) Post Load (0.4ms) SELECT “posts”.* FROM “posts” WHERE “posts”.“id” <= $1 LIMIT $2 [[“id”, 3], [“LIMIT”, 11]]
irb(main):031:0> Post.where(id: …3) Post Load (0.4ms) SELECT “posts”.* FROM “posts” WHERE “posts”.“id” BETWEEN $1 AND $2 LIMIT $3 [[“id”, nil], [“id”, 3], [“LIMIT”, 11]] => #<ActiveRecord::Relation >
``
Neither does it match the behavior of endless ranges which where introduced in 2.6.0.
irb(main):032:0> Post.where(id: 3…) Post Load (0.3ms) SELECT “posts”.* FROM “posts” WHERE “posts”.“id” >= $1 LIMIT $2 [[“id”, 3], [“LIMIT”, 11]] => #<ActiveRecord::Relation >
``
On a beginless range begin is nil, on endless end is nil.
I’m trying to look into this and what could actually be done about it. I’m a bit puzzled at why the endless range works as RangeHandler just lops off the begin and end and passes it to:
predicate_builder.build_bind_attribute
``
Which checks if the value is infinite.
Where along in the chain of RangeHandler → QueryAttribute → ActiveModel::Attribute should this be handled?
