Type error on upsert. by parsonsmatt · Pull Request #885 · yesodweb/persistent

parsonsmatt · 2019-04-11T02:38:52Z

This PR introduces a few type classes that are required on upsert and related functions that have expectations on how many unique keys a record has.

Instances for these classes are generated automatically, and "failing" instances get informative "TypeError"s when you misuse them.

Fixes #868
Fixes #796

Before submitting your PR, check that you've:

Bumped the version number
Documented new APIs with Haddock markup
Added @since declarations to the Haddock

After submitting your PR:

Update the Changelog.md file with a link to your PR
Check that CI passes (or if it fails, for reasons unrelated to your change, like CI timeouts)

parsonsmatt · 2019-04-11T02:45:18Z

Weird, I am getting test suite failures locally for MongoDB:

  src/Init.hs:106:32: 
  1) persistent.putMany handles conflicts by replacing old keys with new records
       expected: [Entity {entityKey = UpsertKey MongoKey {unMongoKey = 5caea9c3d4613542ae00082a}, entityVal = Upsert {upsertEmail = "putMany4", upsertAttr = "new", upsertExtra = "", upsertAge = 2}},Entity {entityKey = UpsertKey MongoKey {unMongoKey = 5caea9c3d4613542ae00082b}, entityVal = Upsert {upsertEmail = "putMany5", upsertAttr = "new", upsertExtra = "", upsertAge = 2}},Entity {entityKey = UpsertKey MongoKey {unMongoKey = 5caea7b6d46135390b00082d}, entityVal = Upsert {upsertEmail = "putMany6", upsertAttr = "new", upsertExtra = "", upsertAge = 2}},Entity {entityKey = UpsertKey MongoKey {unMongoKey = 5caea7b6d46135390b00082c}, entityVal = Upsert {upsertEmail = "putMany7", upsertAttr = "new", upsertExtra = "", upsertAge = 2}}]
        but got: [Entity {entityKey = UpsertKey MongoKey {unMongoKey = 5caea7b6d46135390b00082a}, entityVal = Upsert {upsertEmail = "putMany4", upsertAttr = "new", upsertExtra = "", upsertAge = 2}},Entity {entityKey = UpsertKey MongoKey {unMongoKey = 5caea7b6d46135390b00082b}, entityVal = Upsert {upsertEmail = "putMany5", upsertAttr = "new", upsertExtra = "", upsertAge = 2}},Entity {entityKey = UpsertKey MongoKey {unMongoKey = 5caea7b6d46135390b00082d}, entityVal = Upsert {upsertEmail = "putMany6", upsertAttr = "new", upsertExtra = "", upsertAge = 2}},Entity {entityKey = UpsertKey MongoKey {unMongoKey = 5caea7b6d46135390b00082c}, entityVal = Upsert {upsertEmail = "putMany7", upsertAttr = "new", upsertExtra = "", upsertAge = 2}}]

  To rerun use: --match "/persistent/putMany/handles conflicts by replacing old keys with new records/"

The IDs are different. I did not get these earlier when I was refactoring the test suite, and they don't appear to show up in CI?

parsonsmatt · 2019-04-11T02:50:31Z

persistent-template/Database/Persist/TH.hs

+        $ "Generating Persistent entities now requires the 'UndecidableInstances' "
+        <> "language extension. Please enable it in your file by copy/pasting "
+        <> "this line into the top of your file: \n\n"
+        <> "{-# LANGUAGE UndecidableInstances #-}"


This gives a pretty pleasant error message. The alternative is a bit nasty.

parsonsmatt · 2019-04-11T02:51:28Z

persistent-template/Database/Persist/TH.hs

+
+    singleUniqueKey :: Q [Dec]
+    singleUniqueKey = do
+        expr <- [e|\p -> head (persistUniqueKeysP p)|]


this is safe because we have already guaranteed that there are multiple unique keys

parsonsmatt · 2019-04-11T02:52:29Z

persistent-test/src/PersistentTest.hs

      pBlue30 <- updateGet key25 [PersonAge +=. 2]
      pBlue30 @== Person "Updated" 30 Nothing

-  describe "putMany" $ do


These specs are failing for MongoDB locally for me. The keys are changing, so it isn't updating, but replacing. Strange! I moved them into upsert specs which is pending for now.

parsonsmatt · 2019-04-11T03:49:38Z

persistent-template/Database/Persist/TH.hs

+
+    singleUniqueKey :: Q [Dec]
+    singleUniqueKey = do
+        expr <- [e|\p -> head (persistUniqueKeys p)|]


this is "safe" because the generating code has already guaranteed that we have exactly one unique key.

parsonsmatt · 2019-04-11T03:50:36Z

persistent-template/Database/Persist/TH.hs

+
+    atLeastOneKey :: Q [Dec]
+    atLeastOneKey = do
+        expr <- [e|\p -> NEL.fromList (persistUniqueKeys p)|]


likewise, this is safe because we have a guarantee from the generating code that it has at least one key.

parsonsmatt · 2019-04-11T04:27:27Z

persistent-mongoDB/test/EmbedTestMongo.hs

@@ -1,5 +1,6 @@
 {-# OPTIONS_GHC -fno-warn-unused-binds -fno-warn-orphans -O0 #-}
 {-# LANGUAGE DeriveDataTypeable #-}
+{-# LANGUAGE UndecidableInstances #-}


This change requires UndecidableInstances. I give a nice error message for upgrading.

parsonsmatt · 2019-04-11T04:29:25Z

persistent-test/src/UpsertTest.hs

          Don'tUpdateNull ->
              Nothing
+
+  describe "putMany" $ do


Moved from the PersistentTest module so that mongoDB could skip them for now.

parsonsmatt · 2019-04-11T04:29:39Z

persistent/Database/Persist/Class/PersistUnique.hs

    -- * If the record exists (matched via it's uniqueness constraint), then update the existing record with the parameters which is passed on as list to the function.
    --
-    -- Throws an exception if there is more than 1 uniqueness constraint.
-    --


parsonsmatt · 2019-04-11T04:36:31Z

Do we need to bump the major versions of the component libraries? eg persistent-mysql exports Database.Persist.Sql which has these changes. But it's interface does not change at all, and you need to depend on persistent anyway to use this stuff.

parsonsmatt · 2019-04-11T04:47:12Z

Ah, dang. We can't do TypeError unless we drop support for GHC 7.10.3.

Test suite compiles

41bb51c

parsonsmatt added 5 commits April 10, 2019 20:48

Move putMany into upsert specs

dcc89b5

Actually undefined is wrong

1da4150

Update documentation

456c0f1

update changelogs

dd13ba6

tidy up

11464fe

parsonsmatt commented Apr 11, 2019

View reviewed changes

parsonsmatt added 3 commits April 10, 2019 22:57

Drop 7.10.3 support

9b033dc

clean up postgres test warnings

1cd45ce

Merge branch 'master' into matt/type-error-on-upsert

a1c76a3

parsonsmatt merged commit 04d03e9 into master Apr 11, 2019

parsonsmatt deleted the matt/type-error-on-upsert branch April 11, 2019 21:25

dbaynard mentioned this pull request Jul 15, 2019

cannot cabal install stack-2.1.3 (pantry-0.1.1.0 fails) commercialhaskell/stack#4963

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Type error on upsert.#885

Type error on upsert.#885
parsonsmatt merged 9 commits intomasterfrom
matt/type-error-on-upsert

parsonsmatt commented Apr 11, 2019 •

edited

Loading

Uh oh!

parsonsmatt commented Apr 11, 2019 •

edited

Loading

Uh oh!

parsonsmatt Apr 11, 2019

Uh oh!

parsonsmatt Apr 11, 2019

Uh oh!

parsonsmatt Apr 11, 2019

Uh oh!

parsonsmatt Apr 11, 2019

Uh oh!

parsonsmatt Apr 11, 2019

Uh oh!

parsonsmatt Apr 11, 2019

Uh oh!

parsonsmatt Apr 11, 2019

Uh oh!

parsonsmatt Apr 11, 2019

Uh oh!

parsonsmatt commented Apr 11, 2019

Uh oh!

parsonsmatt commented Apr 11, 2019

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

parsonsmatt commented Apr 11, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

parsonsmatt commented Apr 11, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

parsonsmatt commented Apr 11, 2019

Uh oh!

parsonsmatt commented Apr 11, 2019

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

parsonsmatt commented Apr 11, 2019 •

edited

Loading

parsonsmatt commented Apr 11, 2019 •

edited

Loading