(VB-940) Create ilk history trigger-table #250
Conversation
| VALUES (diff_ilk_identifier, | ||
| new_diff.block_number, | ||
| new_diff.rate, | ||
| ilk_art_before_block(new_diff.ilk_id, new_diff.block_number), |
There was a problem hiding this comment.
Using a similar strategy to our flip/flap/flop trigger-tables, where we insert the prior values of each field aside from the one being provided by the trigger.
Gslaughl
force-pushed
the
vdb-940-ilk-history-table
branch
from
4097110 to
f020017
Compare
|
|
||
|
|
||
| -- +goose StatementBegin | ||
| CREATE OR REPLACE FUNCTION maker.update_time_created() RETURNS TRIGGER |
There was a problem hiding this comment.
created is a little simpler than the rest of the triggers-- we're not adding a new row for each Vat.init, because there will already be a rate diff in the same block, since Vat.init modifies rate.
Also, Vat.init can only be called once per ilk, so created should only be set one time. That's why we're only updating created if it was previously NULL (line 1267)
Gslaughl
requested review from
rmulhol,
elizabethengelman,
m0ar and
yaoandrew
| @@ -0,0 +1,1350 @@ | |||
| -- +goose Up | |||
| CREATE TABLE api.ilk_state_history | |||
There was a problem hiding this comment.
Worth mentioning that exposing this table via postgraphile results in a query called allIlkStateHistories, which is a bad name imo (though nice automatic pluralization by postgraphile!). So I'm kind of leaning towards something like ilk_state_snapshot or historical_ilk_state, in order to get more idiomatic query names
This story populates a table with snapshots of all ilks at every block where they changed, and keeps the history correctly updated if diffs arrive out of order. This story does NOT include re-org safety.