diff options
author | Lauro Caetano <laurocaetano1@gmail.com> | 2014-04-14 21:35:49 -0300 |
---|---|---|
committer | Lauro Caetano <laurocaetano1@gmail.com> | 2014-04-14 22:00:55 -0300 |
commit | a9416a407b3e5256a0a3fbe6c6ee5112ca3b6703 (patch) | |
tree | 72e7606d9d63a2436313d00e594d686ec6bb62e7 /tools | |
parent | 34945e41c24c59268bbde63abfafe20bdc09b775 (diff) | |
download | rails-a9416a407b3e5256a0a3fbe6c6ee5112ca3b6703.tar.gz rails-a9416a407b3e5256a0a3fbe6c6ee5112ca3b6703.tar.bz2 rails-a9416a407b3e5256a0a3fbe6c6ee5112ca3b6703.zip |
Make sure the column_name is different from 'all'.
968c581ea34b5236af14805e6a77913b1cb36238 have fixed the EagerLoadTest, but
not in the correct way.
The problem was when `empty?` or `size` was called on relation. It was
triggering `count(:all)`, which was passing `:all` as the column name to `count`
on Calculations.
On the other hand, the method `calculate` on Calculations was calling
`construct_relation_for_association_calculations` instead of `perform_calculation`,
because `has_include?` was returning `true` since `column_name` was present.
To prevent calling the wrong method to perform the calculation, we have to check
if the `column_name` is present and if it is different from `:all` (which is now used
to correctly do `count` with `select`).
More information here: https://github.com/rails/rails/commit/968c581ea34b5236af14805e6a77913b1cb36238#commitcomment-6006135
Diffstat (limited to 'tools')
0 files changed, 0 insertions, 0 deletions