Add no-downtime migrations support for composite types #21
Conversation
arybczak
force-pushed
the
no-downtime-composites
branch
from
77059bc
to
b81ddb5
Compare
| tableHasLess :: Show t => Text -> t -> Text | ||
| tableHasLess ptype missing = | ||
| "Table in the database has *less*" <+> ptype <+> | ||
| objectHasLess :: Show t => Text -> Text -> t -> Text |
There was a problem hiding this comment.
Minor: ptype seems to always be "columns", maybe these two should be objectHas{Less,More}Columns?
There was a problem hiding this comment.
Let's leave it as-is, I don't think it's a big deal.
| @@ -15,7 +15,7 @@ import Prelude | |||
| data Check = Check { | |||
| chkName :: RawSQL () | |||
| , chkCondition :: RawSQL () | |||
| , chkValidated :: Bool -- ^ Set to 'True' if check is created as NOT VALID and | |||
| , chkValidated :: Bool -- ^ Set to 'False' if check is created as NOT VALID and | |||
| -- not validated afterwards. | |||
There was a problem hiding this comment.
s/not validated/validated/? We're supposed to call sqlValidateCheck at some future point after all.
Maybe rephrase as: "When set to 'False', the check is initially created as NOT VALID. It should be eventually validated with 'sqlValidateCheck'.".
Similarly for fkValidated.
There was a problem hiding this comment.
fkValidated should be set to False if check is created as NOT VALID and left like this (for whatever reason), so the current description is imo alright.
There was a problem hiding this comment.
fkValidated should be set to False if check is created as NOT VALID and left like this (for whatever reason)
OK, maybe change both comments to say this instead of current wording? I find it less confusing than the current one (with the double negative).
|
I pushed additional commit with support for concurrent index creation, please take a look at it. |
|
@23Skidoo Things seem to be good now. Would you mind merging and creating a release? |
|
Merged, will cut a release shortly. |
|
New release is now on Hackage. |
No description provided.