2021-09-22 18:34:53 +03:00
|
|
|
{-# LANGUAGE ApplicativeDo #-}
|
2021-07-15 15:44:26 +03:00
|
|
|
{-# OPTIONS_GHC -fno-warn-orphans #-}
|
|
|
|
|
|
|
|
module Hasura.Backends.MySQL.Instances.Schema where
|
|
|
|
|
server: remove remnants of query plan caching (fix #1795)
Query plan caching was introduced by - I believe - hasura/graphql-engine#1934 in order to reduce the query response latency. During the development of PDV in hasura/graphql-engine#4111, it was found out that the new architecture (for which query plan caching wasn't implemented) performed comparably to the pre-PDV architecture with caching. Hence, it was decided to leave query plan caching until some day in the future when it was deemed necessary.
Well, we're in the future now, and there still isn't a convincing argument for query plan caching. So the time has come to remove some references to query plan caching from the codebase. For the most part, any code being removed would probably not be very well suited to the post-PDV architecture of query execution, so arguably not much is lost.
Apart from simplifying the code, this PR will contribute towards making the GraphQL schema generation more modular, testable, and easier to profile. I'd like to eventually work towards a situation in which it's easy to generate a GraphQL schema parser *in isolation*, without being connected to a database, and then parse a GraphQL query *in isolation*, without even listening any HTTP port. It is important that both of these operations can be examined in detail, and in isolation, since they are two major performance bottlenecks, as well as phases where many important upcoming features hook into.
Implementation
The following have been removed:
- The entirety of `server/src-lib/Hasura/GraphQL/Execute/Plan.hs`
- The core phases of query parsing and execution no longer have any references to query plan caching. Note that this is not to be confused with query *response* caching, which is not affected by this PR. This includes removal of the types:
- - `Opaque`, which is replaced by a tuple. Note that the old implementation was broken and did not adequately hide the constructors.
- - `QueryReusability` (and the `markNotReusable` method). Notably, the implementation of the `ParseT` monad now consists of two, rather than three, monad transformers.
- Cache-related tests (in `server/src-test/Hasura/CacheBoundedSpec.hs`) have been removed .
- References to query plan caching in the documentation.
- The `planCacheOptions` in the `TenantConfig` type class was removed. However, during parsing, unrecognized fields in the YAML config get ignored, so this does not cause a breaking change. (Confirmed manually, as well as in consultation with @sordina.)
- The metrics no longer send cache hit/miss messages.
There are a few places in which one can still find references to query plan caching:
- We still accept the `--query-plan-cache-size` command-line option for backwards compatibility. The `HASURA_QUERY_PLAN_CACHE_SIZE` environment variable is not read.
https://github.com/hasura/graphql-engine-mono/pull/1815
GitOrigin-RevId: 17d92b254ec093c62a7dfeec478658ede0813eb7
2021-07-27 14:51:52 +03:00
|
|
|
import qualified Data.Aeson as J
|
2021-08-04 14:42:24 +03:00
|
|
|
import Data.ByteString (ByteString)
|
server: remove remnants of query plan caching (fix #1795)
Query plan caching was introduced by - I believe - hasura/graphql-engine#1934 in order to reduce the query response latency. During the development of PDV in hasura/graphql-engine#4111, it was found out that the new architecture (for which query plan caching wasn't implemented) performed comparably to the pre-PDV architecture with caching. Hence, it was decided to leave query plan caching until some day in the future when it was deemed necessary.
Well, we're in the future now, and there still isn't a convincing argument for query plan caching. So the time has come to remove some references to query plan caching from the codebase. For the most part, any code being removed would probably not be very well suited to the post-PDV architecture of query execution, so arguably not much is lost.
Apart from simplifying the code, this PR will contribute towards making the GraphQL schema generation more modular, testable, and easier to profile. I'd like to eventually work towards a situation in which it's easy to generate a GraphQL schema parser *in isolation*, without being connected to a database, and then parse a GraphQL query *in isolation*, without even listening any HTTP port. It is important that both of these operations can be examined in detail, and in isolation, since they are two major performance bottlenecks, as well as phases where many important upcoming features hook into.
Implementation
The following have been removed:
- The entirety of `server/src-lib/Hasura/GraphQL/Execute/Plan.hs`
- The core phases of query parsing and execution no longer have any references to query plan caching. Note that this is not to be confused with query *response* caching, which is not affected by this PR. This includes removal of the types:
- - `Opaque`, which is replaced by a tuple. Note that the old implementation was broken and did not adequately hide the constructors.
- - `QueryReusability` (and the `markNotReusable` method). Notably, the implementation of the `ParseT` monad now consists of two, rather than three, monad transformers.
- Cache-related tests (in `server/src-test/Hasura/CacheBoundedSpec.hs`) have been removed .
- References to query plan caching in the documentation.
- The `planCacheOptions` in the `TenantConfig` type class was removed. However, during parsing, unrecognized fields in the YAML config get ignored, so this does not cause a breaking change. (Confirmed manually, as well as in consultation with @sordina.)
- The metrics no longer send cache hit/miss messages.
There are a few places in which one can still find references to query plan caching:
- We still accept the `--query-plan-cache-size` command-line option for backwards compatibility. The `HASURA_QUERY_PLAN_CACHE_SIZE` environment variable is not read.
https://github.com/hasura/graphql-engine-mono/pull/1815
GitOrigin-RevId: 17d92b254ec093c62a7dfeec478658ede0813eb7
2021-07-27 14:51:52 +03:00
|
|
|
import qualified Data.HashMap.Strict as HM
|
|
|
|
import qualified Data.List.NonEmpty as NE
|
2021-08-04 14:42:24 +03:00
|
|
|
import Data.Text.Encoding (encodeUtf8)
|
2021-07-15 15:44:26 +03:00
|
|
|
import Data.Text.Extended
|
server: remove remnants of query plan caching (fix #1795)
Query plan caching was introduced by - I believe - hasura/graphql-engine#1934 in order to reduce the query response latency. During the development of PDV in hasura/graphql-engine#4111, it was found out that the new architecture (for which query plan caching wasn't implemented) performed comparably to the pre-PDV architecture with caching. Hence, it was decided to leave query plan caching until some day in the future when it was deemed necessary.
Well, we're in the future now, and there still isn't a convincing argument for query plan caching. So the time has come to remove some references to query plan caching from the codebase. For the most part, any code being removed would probably not be very well suited to the post-PDV architecture of query execution, so arguably not much is lost.
Apart from simplifying the code, this PR will contribute towards making the GraphQL schema generation more modular, testable, and easier to profile. I'd like to eventually work towards a situation in which it's easy to generate a GraphQL schema parser *in isolation*, without being connected to a database, and then parse a GraphQL query *in isolation*, without even listening any HTTP port. It is important that both of these operations can be examined in detail, and in isolation, since they are two major performance bottlenecks, as well as phases where many important upcoming features hook into.
Implementation
The following have been removed:
- The entirety of `server/src-lib/Hasura/GraphQL/Execute/Plan.hs`
- The core phases of query parsing and execution no longer have any references to query plan caching. Note that this is not to be confused with query *response* caching, which is not affected by this PR. This includes removal of the types:
- - `Opaque`, which is replaced by a tuple. Note that the old implementation was broken and did not adequately hide the constructors.
- - `QueryReusability` (and the `markNotReusable` method). Notably, the implementation of the `ParseT` monad now consists of two, rather than three, monad transformers.
- Cache-related tests (in `server/src-test/Hasura/CacheBoundedSpec.hs`) have been removed .
- References to query plan caching in the documentation.
- The `planCacheOptions` in the `TenantConfig` type class was removed. However, during parsing, unrecognized fields in the YAML config get ignored, so this does not cause a breaking change. (Confirmed manually, as well as in consultation with @sordina.)
- The metrics no longer send cache hit/miss messages.
There are a few places in which one can still find references to query plan caching:
- We still accept the `--query-plan-cache-size` command-line option for backwards compatibility. The `HASURA_QUERY_PLAN_CACHE_SIZE` environment variable is not read.
https://github.com/hasura/graphql-engine-mono/pull/1815
GitOrigin-RevId: 17d92b254ec093c62a7dfeec478658ede0813eb7
2021-07-27 14:51:52 +03:00
|
|
|
import qualified Database.MySQL.Base.Types as MySQL
|
|
|
|
import qualified Hasura.Backends.MySQL.Types as MySQL
|
2021-07-15 15:44:26 +03:00
|
|
|
import Hasura.Base.Error
|
server: remove remnants of query plan caching (fix #1795)
Query plan caching was introduced by - I believe - hasura/graphql-engine#1934 in order to reduce the query response latency. During the development of PDV in hasura/graphql-engine#4111, it was found out that the new architecture (for which query plan caching wasn't implemented) performed comparably to the pre-PDV architecture with caching. Hence, it was decided to leave query plan caching until some day in the future when it was deemed necessary.
Well, we're in the future now, and there still isn't a convincing argument for query plan caching. So the time has come to remove some references to query plan caching from the codebase. For the most part, any code being removed would probably not be very well suited to the post-PDV architecture of query execution, so arguably not much is lost.
Apart from simplifying the code, this PR will contribute towards making the GraphQL schema generation more modular, testable, and easier to profile. I'd like to eventually work towards a situation in which it's easy to generate a GraphQL schema parser *in isolation*, without being connected to a database, and then parse a GraphQL query *in isolation*, without even listening any HTTP port. It is important that both of these operations can be examined in detail, and in isolation, since they are two major performance bottlenecks, as well as phases where many important upcoming features hook into.
Implementation
The following have been removed:
- The entirety of `server/src-lib/Hasura/GraphQL/Execute/Plan.hs`
- The core phases of query parsing and execution no longer have any references to query plan caching. Note that this is not to be confused with query *response* caching, which is not affected by this PR. This includes removal of the types:
- - `Opaque`, which is replaced by a tuple. Note that the old implementation was broken and did not adequately hide the constructors.
- - `QueryReusability` (and the `markNotReusable` method). Notably, the implementation of the `ParseT` monad now consists of two, rather than three, monad transformers.
- Cache-related tests (in `server/src-test/Hasura/CacheBoundedSpec.hs`) have been removed .
- References to query plan caching in the documentation.
- The `planCacheOptions` in the `TenantConfig` type class was removed. However, during parsing, unrecognized fields in the YAML config get ignored, so this does not cause a breaking change. (Confirmed manually, as well as in consultation with @sordina.)
- The metrics no longer send cache hit/miss messages.
There are a few places in which one can still find references to query plan caching:
- We still accept the `--query-plan-cache-size` command-line option for backwards compatibility. The `HASURA_QUERY_PLAN_CACHE_SIZE` environment variable is not read.
https://github.com/hasura/graphql-engine-mono/pull/1815
GitOrigin-RevId: 17d92b254ec093c62a7dfeec478658ede0813eb7
2021-07-27 14:51:52 +03:00
|
|
|
import Hasura.GraphQL.Parser hiding (EnumValueInfo, field)
|
|
|
|
import qualified Hasura.GraphQL.Parser as P
|
|
|
|
import Hasura.GraphQL.Parser.Internal.Parser hiding (field)
|
2021-07-15 15:44:26 +03:00
|
|
|
import Hasura.GraphQL.Schema.Backend
|
server: remove remnants of query plan caching (fix #1795)
Query plan caching was introduced by - I believe - hasura/graphql-engine#1934 in order to reduce the query response latency. During the development of PDV in hasura/graphql-engine#4111, it was found out that the new architecture (for which query plan caching wasn't implemented) performed comparably to the pre-PDV architecture with caching. Hence, it was decided to leave query plan caching until some day in the future when it was deemed necessary.
Well, we're in the future now, and there still isn't a convincing argument for query plan caching. So the time has come to remove some references to query plan caching from the codebase. For the most part, any code being removed would probably not be very well suited to the post-PDV architecture of query execution, so arguably not much is lost.
Apart from simplifying the code, this PR will contribute towards making the GraphQL schema generation more modular, testable, and easier to profile. I'd like to eventually work towards a situation in which it's easy to generate a GraphQL schema parser *in isolation*, without being connected to a database, and then parse a GraphQL query *in isolation*, without even listening any HTTP port. It is important that both of these operations can be examined in detail, and in isolation, since they are two major performance bottlenecks, as well as phases where many important upcoming features hook into.
Implementation
The following have been removed:
- The entirety of `server/src-lib/Hasura/GraphQL/Execute/Plan.hs`
- The core phases of query parsing and execution no longer have any references to query plan caching. Note that this is not to be confused with query *response* caching, which is not affected by this PR. This includes removal of the types:
- - `Opaque`, which is replaced by a tuple. Note that the old implementation was broken and did not adequately hide the constructors.
- - `QueryReusability` (and the `markNotReusable` method). Notably, the implementation of the `ParseT` monad now consists of two, rather than three, monad transformers.
- Cache-related tests (in `server/src-test/Hasura/CacheBoundedSpec.hs`) have been removed .
- References to query plan caching in the documentation.
- The `planCacheOptions` in the `TenantConfig` type class was removed. However, during parsing, unrecognized fields in the YAML config get ignored, so this does not cause a breaking change. (Confirmed manually, as well as in consultation with @sordina.)
- The metrics no longer send cache hit/miss messages.
There are a few places in which one can still find references to query plan caching:
- We still accept the `--query-plan-cache-size` command-line option for backwards compatibility. The `HASURA_QUERY_PLAN_CACHE_SIZE` environment variable is not read.
https://github.com/hasura/graphql-engine-mono/pull/1815
GitOrigin-RevId: 17d92b254ec093c62a7dfeec478658ede0813eb7
2021-07-27 14:51:52 +03:00
|
|
|
import qualified Hasura.GraphQL.Schema.Build as GSB
|
2021-07-15 15:44:26 +03:00
|
|
|
import Hasura.GraphQL.Schema.Select
|
|
|
|
import Hasura.Prelude
|
|
|
|
import Hasura.RQL.IR
|
server: remove remnants of query plan caching (fix #1795)
Query plan caching was introduced by - I believe - hasura/graphql-engine#1934 in order to reduce the query response latency. During the development of PDV in hasura/graphql-engine#4111, it was found out that the new architecture (for which query plan caching wasn't implemented) performed comparably to the pre-PDV architecture with caching. Hence, it was decided to leave query plan caching until some day in the future when it was deemed necessary.
Well, we're in the future now, and there still isn't a convincing argument for query plan caching. So the time has come to remove some references to query plan caching from the codebase. For the most part, any code being removed would probably not be very well suited to the post-PDV architecture of query execution, so arguably not much is lost.
Apart from simplifying the code, this PR will contribute towards making the GraphQL schema generation more modular, testable, and easier to profile. I'd like to eventually work towards a situation in which it's easy to generate a GraphQL schema parser *in isolation*, without being connected to a database, and then parse a GraphQL query *in isolation*, without even listening any HTTP port. It is important that both of these operations can be examined in detail, and in isolation, since they are two major performance bottlenecks, as well as phases where many important upcoming features hook into.
Implementation
The following have been removed:
- The entirety of `server/src-lib/Hasura/GraphQL/Execute/Plan.hs`
- The core phases of query parsing and execution no longer have any references to query plan caching. Note that this is not to be confused with query *response* caching, which is not affected by this PR. This includes removal of the types:
- - `Opaque`, which is replaced by a tuple. Note that the old implementation was broken and did not adequately hide the constructors.
- - `QueryReusability` (and the `markNotReusable` method). Notably, the implementation of the `ParseT` monad now consists of two, rather than three, monad transformers.
- Cache-related tests (in `server/src-test/Hasura/CacheBoundedSpec.hs`) have been removed .
- References to query plan caching in the documentation.
- The `planCacheOptions` in the `TenantConfig` type class was removed. However, during parsing, unrecognized fields in the YAML config get ignored, so this does not cause a breaking change. (Confirmed manually, as well as in consultation with @sordina.)
- The metrics no longer send cache hit/miss messages.
There are a few places in which one can still find references to query plan caching:
- We still accept the `--query-plan-cache-size` command-line option for backwards compatibility. The `HASURA_QUERY_PLAN_CACHE_SIZE` environment variable is not read.
https://github.com/hasura/graphql-engine-mono/pull/1815
GitOrigin-RevId: 17d92b254ec093c62a7dfeec478658ede0813eb7
2021-07-27 14:51:52 +03:00
|
|
|
import qualified Hasura.RQL.IR.Select as IR
|
|
|
|
import qualified Hasura.RQL.IR.Update as IR
|
|
|
|
import Hasura.RQL.Types as RQL
|
|
|
|
import qualified Language.GraphQL.Draft.Syntax as G
|
2021-07-15 15:44:26 +03:00
|
|
|
|
|
|
|
instance BackendSchema 'MySQL where
|
|
|
|
buildTableQueryFields = GSB.buildTableQueryFields
|
|
|
|
buildTableRelayQueryFields = buildTableRelayQueryFields'
|
|
|
|
buildTableInsertMutationFields = buildTableInsertMutationFields'
|
|
|
|
buildTableUpdateMutationFields = buildTableUpdateMutationFields'
|
|
|
|
buildTableDeleteMutationFields = buildTableDeleteMutationFields'
|
|
|
|
buildFunctionQueryFields = buildFunctionQueryFields'
|
|
|
|
buildFunctionRelayQueryFields = buildFunctionRelayQueryFields'
|
|
|
|
buildFunctionMutationFields = buildFunctionMutationFields'
|
|
|
|
relayExtension = Nothing
|
|
|
|
tableArguments = mysqlTableArgs
|
|
|
|
nodesAggExtension = Just ()
|
|
|
|
columnParser = columnParser'
|
|
|
|
jsonPathArg = jsonPathArg'
|
|
|
|
orderByOperators = orderByOperators'
|
|
|
|
comparisonExps = comparisonExps'
|
|
|
|
updateOperators = updateOperators'
|
2021-08-04 14:42:24 +03:00
|
|
|
mkCountType = error "mkCountType: MySQL backend does not support this operation yet."
|
|
|
|
aggregateOrderByCountType = error "aggregateOrderByCountType: MySQL backend does not support this operation yet."
|
|
|
|
computedField = error "computedField: MySQL backend does not support this operation yet."
|
|
|
|
node = error "node: MySQL backend does not support this operation yet."
|
|
|
|
columnDefaultValue = error "columnDefaultValue: MySQL backend does not support this operation yet."
|
2021-08-31 16:34:43 +03:00
|
|
|
nestedInsertsExtension = Nothing
|
2021-07-15 15:44:26 +03:00
|
|
|
|
|
|
|
mysqlTableArgs
|
|
|
|
:: forall r m n
|
|
|
|
. MonadBuildSchema 'MySQL r m n
|
|
|
|
=> SourceName
|
|
|
|
-> TableInfo 'MySQL
|
|
|
|
-> SelPermInfo 'MySQL
|
|
|
|
-> m (InputFieldsParser n (IR.SelectArgsG 'MySQL (UnpreparedValue 'MySQL)))
|
|
|
|
mysqlTableArgs sourceName tableInfo selectPermissions = do
|
|
|
|
whereParser <- tableWhereArg sourceName tableInfo selectPermissions
|
|
|
|
orderByParser <- tableOrderByArg sourceName tableInfo selectPermissions
|
|
|
|
pure do
|
|
|
|
whereArg <- whereParser
|
|
|
|
orderByArg <- orderByParser
|
|
|
|
limitArg <- tableLimitArg
|
|
|
|
offsetArg <- tableOffsetArg
|
|
|
|
pure $ IR.SelectArgs
|
|
|
|
{ IR._saWhere = whereArg
|
|
|
|
, IR._saOrderBy = orderByArg
|
|
|
|
, IR._saLimit = limitArg
|
|
|
|
, IR._saOffset = offsetArg
|
|
|
|
, IR._saDistinct = Nothing
|
|
|
|
}
|
|
|
|
|
|
|
|
buildTableRelayQueryFields' ::
|
|
|
|
MonadBuildSchema 'MySQL r m n =>
|
|
|
|
SourceName ->
|
|
|
|
RQL.SourceConfig 'MySQL ->
|
2021-09-23 15:37:56 +03:00
|
|
|
Maybe RQL.QueryTagsConfig ->
|
2021-07-15 15:44:26 +03:00
|
|
|
RQL.TableName 'MySQL ->
|
|
|
|
TableInfo 'MySQL ->
|
|
|
|
G.Name ->
|
|
|
|
NESeq (ColumnInfo 'MySQL) ->
|
|
|
|
SelPermInfo 'MySQL ->
|
2021-07-26 16:03:51 +03:00
|
|
|
m [FieldParser n (QueryRootField UnpreparedValue)]
|
2021-09-23 15:37:56 +03:00
|
|
|
buildTableRelayQueryFields' _sourceName _sourceInfo _queryTagsConfig _tableName _tableInfo _gqlName _pkeyColumns _selPerms =
|
2021-07-15 15:44:26 +03:00
|
|
|
pure []
|
|
|
|
|
|
|
|
|
|
|
|
buildTableInsertMutationFields' ::
|
|
|
|
MonadBuildSchema 'MySQL r m n =>
|
|
|
|
SourceName ->
|
|
|
|
RQL.SourceConfig 'MySQL ->
|
2021-09-23 15:37:56 +03:00
|
|
|
Maybe RQL.QueryTagsConfig ->
|
2021-07-15 15:44:26 +03:00
|
|
|
RQL.TableName 'MySQL ->
|
|
|
|
TableInfo 'MySQL ->
|
|
|
|
G.Name ->
|
|
|
|
InsPermInfo 'MySQL ->
|
|
|
|
Maybe (SelPermInfo 'MySQL) ->
|
|
|
|
Maybe (UpdPermInfo 'MySQL) ->
|
2021-07-26 16:03:51 +03:00
|
|
|
m [FieldParser n (MutationRootField UnpreparedValue)]
|
2021-09-23 15:37:56 +03:00
|
|
|
buildTableInsertMutationFields' _sourceName _sourceInfo _queryTagsConfig _tableName _tableInfo _gqlName _insPerms _selPerms _updPerms =
|
2021-07-15 15:44:26 +03:00
|
|
|
pure []
|
|
|
|
|
|
|
|
|
|
|
|
buildTableUpdateMutationFields' ::
|
|
|
|
MonadBuildSchema 'MySQL r m n =>
|
|
|
|
SourceName ->
|
|
|
|
RQL.SourceConfig 'MySQL ->
|
2021-09-23 15:37:56 +03:00
|
|
|
Maybe RQL.QueryTagsConfig ->
|
2021-07-15 15:44:26 +03:00
|
|
|
RQL.TableName 'MySQL ->
|
|
|
|
TableInfo 'MySQL ->
|
|
|
|
G.Name ->
|
|
|
|
UpdPermInfo 'MySQL ->
|
|
|
|
Maybe (SelPermInfo 'MySQL) ->
|
2021-07-26 16:03:51 +03:00
|
|
|
m [FieldParser n (MutationRootField UnpreparedValue)]
|
2021-09-23 15:37:56 +03:00
|
|
|
buildTableUpdateMutationFields' _sourceName _sourceInfo _queryTagsConfig _tableName _tableInfo _gqlName _updPerns _selPerms =
|
2021-07-15 15:44:26 +03:00
|
|
|
pure []
|
|
|
|
|
|
|
|
|
|
|
|
buildTableDeleteMutationFields' ::
|
|
|
|
MonadBuildSchema 'MySQL r m n =>
|
|
|
|
SourceName ->
|
|
|
|
RQL.SourceConfig 'MySQL ->
|
2021-09-23 15:37:56 +03:00
|
|
|
Maybe RQL.QueryTagsConfig ->
|
2021-07-15 15:44:26 +03:00
|
|
|
RQL.TableName 'MySQL ->
|
|
|
|
TableInfo 'MySQL ->
|
|
|
|
G.Name ->
|
|
|
|
DelPermInfo 'MySQL ->
|
|
|
|
Maybe (SelPermInfo 'MySQL) ->
|
2021-07-26 16:03:51 +03:00
|
|
|
m [FieldParser n (MutationRootField UnpreparedValue)]
|
2021-09-23 15:37:56 +03:00
|
|
|
buildTableDeleteMutationFields' _sourceName _sourceInfo _queryTagsConfig _tableName _tableInfo _gqlName _delPerns _selPerms =
|
2021-07-15 15:44:26 +03:00
|
|
|
pure []
|
|
|
|
|
|
|
|
|
|
|
|
buildFunctionQueryFields' ::
|
|
|
|
MonadBuildSchema 'MySQL r m n =>
|
|
|
|
SourceName ->
|
|
|
|
RQL.SourceConfig 'MySQL ->
|
2021-09-23 15:37:56 +03:00
|
|
|
Maybe RQL.QueryTagsConfig ->
|
2021-07-15 15:44:26 +03:00
|
|
|
FunctionName 'MySQL ->
|
|
|
|
FunctionInfo 'MySQL ->
|
|
|
|
RQL.TableName 'MySQL ->
|
|
|
|
SelPermInfo 'MySQL ->
|
2021-07-26 16:03:51 +03:00
|
|
|
m [FieldParser n (QueryRootField UnpreparedValue)]
|
2021-09-23 15:37:56 +03:00
|
|
|
buildFunctionQueryFields' _ _ _ _ _ _ _ =
|
2021-07-15 15:44:26 +03:00
|
|
|
pure []
|
|
|
|
|
|
|
|
|
|
|
|
buildFunctionRelayQueryFields' ::
|
|
|
|
MonadBuildSchema 'MySQL r m n =>
|
|
|
|
SourceName ->
|
|
|
|
RQL.SourceConfig 'MySQL ->
|
2021-09-23 15:37:56 +03:00
|
|
|
Maybe RQL.QueryTagsConfig ->
|
2021-07-15 15:44:26 +03:00
|
|
|
FunctionName 'MySQL ->
|
|
|
|
FunctionInfo 'MySQL ->
|
|
|
|
RQL.TableName 'MySQL ->
|
|
|
|
NESeq (ColumnInfo 'MySQL) ->
|
|
|
|
SelPermInfo 'MySQL ->
|
2021-07-26 16:03:51 +03:00
|
|
|
m [FieldParser n (QueryRootField UnpreparedValue)]
|
2021-09-23 15:37:56 +03:00
|
|
|
buildFunctionRelayQueryFields' _sourceName _sourceInfo _queryTagsConfig _functionName _functionInfo _tableName _pkeyColumns _selPerms =
|
2021-07-15 15:44:26 +03:00
|
|
|
pure []
|
|
|
|
|
|
|
|
|
|
|
|
buildFunctionMutationFields' ::
|
|
|
|
MonadBuildSchema 'MySQL r m n =>
|
|
|
|
SourceName ->
|
|
|
|
RQL.SourceConfig 'MySQL ->
|
2021-09-23 15:37:56 +03:00
|
|
|
Maybe RQL.QueryTagsConfig ->
|
2021-07-15 15:44:26 +03:00
|
|
|
FunctionName 'MySQL ->
|
|
|
|
FunctionInfo 'MySQL ->
|
|
|
|
RQL.TableName 'MySQL ->
|
|
|
|
SelPermInfo 'MySQL ->
|
2021-07-26 16:03:51 +03:00
|
|
|
m [FieldParser n (MutationRootField UnpreparedValue)]
|
2021-09-23 15:37:56 +03:00
|
|
|
buildFunctionMutationFields' _ _ _ _ _ _ _ =
|
2021-07-15 15:44:26 +03:00
|
|
|
pure []
|
|
|
|
|
2021-08-04 14:42:24 +03:00
|
|
|
bsParser :: MonadParse m => Parser 'Both m ByteString
|
|
|
|
bsParser = encodeUtf8 <$> P.string
|
2021-07-15 15:44:26 +03:00
|
|
|
|
|
|
|
columnParser' :: (MonadSchema n m, MonadError QErr m) =>
|
|
|
|
ColumnType 'MySQL ->
|
|
|
|
G.Nullability ->
|
server: remove remnants of query plan caching (fix #1795)
Query plan caching was introduced by - I believe - hasura/graphql-engine#1934 in order to reduce the query response latency. During the development of PDV in hasura/graphql-engine#4111, it was found out that the new architecture (for which query plan caching wasn't implemented) performed comparably to the pre-PDV architecture with caching. Hence, it was decided to leave query plan caching until some day in the future when it was deemed necessary.
Well, we're in the future now, and there still isn't a convincing argument for query plan caching. So the time has come to remove some references to query plan caching from the codebase. For the most part, any code being removed would probably not be very well suited to the post-PDV architecture of query execution, so arguably not much is lost.
Apart from simplifying the code, this PR will contribute towards making the GraphQL schema generation more modular, testable, and easier to profile. I'd like to eventually work towards a situation in which it's easy to generate a GraphQL schema parser *in isolation*, without being connected to a database, and then parse a GraphQL query *in isolation*, without even listening any HTTP port. It is important that both of these operations can be examined in detail, and in isolation, since they are two major performance bottlenecks, as well as phases where many important upcoming features hook into.
Implementation
The following have been removed:
- The entirety of `server/src-lib/Hasura/GraphQL/Execute/Plan.hs`
- The core phases of query parsing and execution no longer have any references to query plan caching. Note that this is not to be confused with query *response* caching, which is not affected by this PR. This includes removal of the types:
- - `Opaque`, which is replaced by a tuple. Note that the old implementation was broken and did not adequately hide the constructors.
- - `QueryReusability` (and the `markNotReusable` method). Notably, the implementation of the `ParseT` monad now consists of two, rather than three, monad transformers.
- Cache-related tests (in `server/src-test/Hasura/CacheBoundedSpec.hs`) have been removed .
- References to query plan caching in the documentation.
- The `planCacheOptions` in the `TenantConfig` type class was removed. However, during parsing, unrecognized fields in the YAML config get ignored, so this does not cause a breaking change. (Confirmed manually, as well as in consultation with @sordina.)
- The metrics no longer send cache hit/miss messages.
There are a few places in which one can still find references to query plan caching:
- We still accept the `--query-plan-cache-size` command-line option for backwards compatibility. The `HASURA_QUERY_PLAN_CACHE_SIZE` environment variable is not read.
https://github.com/hasura/graphql-engine-mono/pull/1815
GitOrigin-RevId: 17d92b254ec093c62a7dfeec478658ede0813eb7
2021-07-27 14:51:52 +03:00
|
|
|
m (Parser 'Both n (ValueWithOrigin (ColumnValue 'MySQL)))
|
2021-07-15 15:44:26 +03:00
|
|
|
columnParser' columnType (G.Nullability isNullable) =
|
server: remove remnants of query plan caching (fix #1795)
Query plan caching was introduced by - I believe - hasura/graphql-engine#1934 in order to reduce the query response latency. During the development of PDV in hasura/graphql-engine#4111, it was found out that the new architecture (for which query plan caching wasn't implemented) performed comparably to the pre-PDV architecture with caching. Hence, it was decided to leave query plan caching until some day in the future when it was deemed necessary.
Well, we're in the future now, and there still isn't a convincing argument for query plan caching. So the time has come to remove some references to query plan caching from the codebase. For the most part, any code being removed would probably not be very well suited to the post-PDV architecture of query execution, so arguably not much is lost.
Apart from simplifying the code, this PR will contribute towards making the GraphQL schema generation more modular, testable, and easier to profile. I'd like to eventually work towards a situation in which it's easy to generate a GraphQL schema parser *in isolation*, without being connected to a database, and then parse a GraphQL query *in isolation*, without even listening any HTTP port. It is important that both of these operations can be examined in detail, and in isolation, since they are two major performance bottlenecks, as well as phases where many important upcoming features hook into.
Implementation
The following have been removed:
- The entirety of `server/src-lib/Hasura/GraphQL/Execute/Plan.hs`
- The core phases of query parsing and execution no longer have any references to query plan caching. Note that this is not to be confused with query *response* caching, which is not affected by this PR. This includes removal of the types:
- - `Opaque`, which is replaced by a tuple. Note that the old implementation was broken and did not adequately hide the constructors.
- - `QueryReusability` (and the `markNotReusable` method). Notably, the implementation of the `ParseT` monad now consists of two, rather than three, monad transformers.
- Cache-related tests (in `server/src-test/Hasura/CacheBoundedSpec.hs`) have been removed .
- References to query plan caching in the documentation.
- The `planCacheOptions` in the `TenantConfig` type class was removed. However, during parsing, unrecognized fields in the YAML config get ignored, so this does not cause a breaking change. (Confirmed manually, as well as in consultation with @sordina.)
- The metrics no longer send cache hit/miss messages.
There are a few places in which one can still find references to query plan caching:
- We still accept the `--query-plan-cache-size` command-line option for backwards compatibility. The `HASURA_QUERY_PLAN_CACHE_SIZE` environment variable is not read.
https://github.com/hasura/graphql-engine-mono/pull/1815
GitOrigin-RevId: 17d92b254ec093c62a7dfeec478658ede0813eb7
2021-07-27 14:51:52 +03:00
|
|
|
peelWithOrigin . fmap (ColumnValue columnType) <$> case columnType of
|
2021-07-15 15:44:26 +03:00
|
|
|
ColumnScalar scalarType -> case scalarType of
|
|
|
|
MySQL.Decimal -> pure $ possiblyNullable scalarType $ MySQL.DecimalValue <$> P.float
|
2021-08-04 14:42:24 +03:00
|
|
|
MySQL.Tiny -> pure $ possiblyNullable scalarType $ MySQL.TinyValue <$> P.int
|
|
|
|
MySQL.Short -> pure $ possiblyNullable scalarType $ MySQL.SmallValue <$> P.int
|
|
|
|
MySQL.Long -> pure $ possiblyNullable scalarType $ MySQL.IntValue <$> P.int
|
2021-07-15 15:44:26 +03:00
|
|
|
MySQL.Float -> pure $ possiblyNullable scalarType $ MySQL.FloatValue <$> P.float
|
2021-08-04 14:42:24 +03:00
|
|
|
MySQL.Double -> pure $ possiblyNullable scalarType $ MySQL.DoubleValue <$> P.float
|
|
|
|
MySQL.Null -> pure $ possiblyNullable scalarType $ MySQL.NullValue <$ P.string
|
2021-07-15 15:44:26 +03:00
|
|
|
MySQL.LongLong -> pure $ possiblyNullable scalarType $ MySQL.BigValue <$> P.int
|
2021-08-04 14:42:24 +03:00
|
|
|
MySQL.Int24 -> pure $ possiblyNullable scalarType $ MySQL.MediumValue <$> P.int
|
|
|
|
MySQL.Date -> pure $ possiblyNullable scalarType $ MySQL.DateValue <$> P.string
|
|
|
|
MySQL.Year -> pure $ possiblyNullable scalarType $ MySQL.YearValue <$> P.string
|
|
|
|
MySQL.Bit -> pure $ possiblyNullable scalarType $ MySQL.BitValue <$> P.boolean
|
|
|
|
MySQL.String -> pure $ possiblyNullable scalarType $ MySQL.VarcharValue <$> P.string
|
|
|
|
MySQL.VarChar -> pure $ possiblyNullable scalarType $ MySQL.VarcharValue <$> P.string
|
|
|
|
MySQL.DateTime -> pure $ possiblyNullable scalarType $ MySQL.DatetimeValue <$> P.string
|
|
|
|
MySQL.Blob -> pure $ possiblyNullable scalarType $ MySQL.BlobValue <$> bsParser
|
|
|
|
MySQL.Timestamp -> pure $ possiblyNullable scalarType $ MySQL.TimestampValue <$> P.string
|
|
|
|
_ -> do
|
|
|
|
name <- MySQL.mkMySQLScalarTypeName scalarType
|
|
|
|
let schemaType = P.NonNullable $ P.TNamed $ P.mkDefinition name Nothing P.TIScalar
|
|
|
|
pure $ Parser
|
|
|
|
{ pType = schemaType
|
|
|
|
, pParser =
|
|
|
|
valueToJSON (P.toGraphQLType schemaType) >=>
|
|
|
|
either (parseErrorWith ParseFailed . qeError) pure . (MySQL.parseScalarValue scalarType)
|
|
|
|
}
|
2021-07-15 15:44:26 +03:00
|
|
|
ColumnEnumReference (EnumReference tableName enumValues) ->
|
|
|
|
case nonEmpty (HM.toList enumValues) of
|
|
|
|
Just enumValuesList -> do
|
|
|
|
tableGQLName <- tableGraphQLName @'MySQL tableName `onLeft` throwError
|
|
|
|
let enumName = tableGQLName <> $$(G.litName "_enum")
|
|
|
|
pure $ possiblyNullable MySQL.VarChar $ P.enum enumName Nothing (mkEnumValue <$> enumValuesList)
|
|
|
|
Nothing -> throw400 ValidationFailed "empty enum values"
|
|
|
|
where
|
|
|
|
possiblyNullable :: (MonadParse m) => MySQL.Type -> Parser 'Both m MySQL.ScalarValue -> Parser 'Both m MySQL.ScalarValue
|
|
|
|
possiblyNullable _scalarType
|
|
|
|
| isNullable = fmap (fromMaybe MySQL.NullValue) . P.nullable
|
|
|
|
| otherwise = id
|
|
|
|
mkEnumValue :: (EnumValue, EnumValueInfo) -> (P.Definition P.EnumValueInfo, RQL.ScalarValue 'MySQL)
|
|
|
|
mkEnumValue (RQL.EnumValue value, EnumValueInfo description) =
|
|
|
|
( P.mkDefinition value (G.Description <$> description) P.EnumValueInfo
|
|
|
|
, MySQL.VarcharValue $ G.unName value
|
|
|
|
)
|
|
|
|
throughJSON scalarName =
|
|
|
|
let schemaType = P.NonNullable $ P.TNamed $ P.mkDefinition scalarName Nothing P.TIScalar
|
|
|
|
in Parser
|
|
|
|
{ pType = schemaType
|
|
|
|
, pParser =
|
|
|
|
valueToJSON (P.toGraphQLType schemaType) >=>
|
|
|
|
either (parseErrorWith ParseFailed . qeError) pure . runAesonParser J.parseJSON
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
jsonPathArg' ::
|
|
|
|
MonadParse n =>
|
|
|
|
ColumnType 'MySQL ->
|
|
|
|
InputFieldsParser n (Maybe (IR.ColumnOp 'MySQL))
|
|
|
|
jsonPathArg' _columnType = pure Nothing
|
|
|
|
|
|
|
|
|
|
|
|
orderByOperators' :: NonEmpty (Definition P.EnumValueInfo, (BasicOrderType 'MySQL, NullsOrderType 'MySQL))
|
|
|
|
orderByOperators' =
|
|
|
|
NE.fromList
|
|
|
|
[ ( define $$(G.litName "asc") "in ascending order, nulls first"
|
|
|
|
, (MySQL.Asc, MySQL.NullsFirst)
|
|
|
|
)
|
|
|
|
, ( define $$(G.litName "asc_nulls_first") "in ascending order, nulls first"
|
|
|
|
, (MySQL.Asc, MySQL.NullsFirst)
|
|
|
|
)
|
|
|
|
, ( define $$(G.litName "asc_nulls_last") "in ascending order, nulls last"
|
|
|
|
, (MySQL.Asc, MySQL.NullsLast)
|
|
|
|
)
|
|
|
|
, ( define $$(G.litName "desc") "in descending order, nulls last"
|
|
|
|
, (MySQL.Desc, MySQL.NullsLast)
|
|
|
|
)
|
|
|
|
, ( define $$(G.litName "desc_nulls_first") "in descending order, nulls first"
|
|
|
|
, (MySQL.Desc, MySQL.NullsFirst)
|
|
|
|
)
|
|
|
|
, ( define $$(G.litName "desc_nulls_last") "in descending order, nulls last"
|
|
|
|
, (MySQL.Desc, MySQL.NullsLast)
|
|
|
|
)
|
|
|
|
]
|
|
|
|
where
|
|
|
|
define name desc = P.mkDefinition name (Just desc) P.EnumValueInfo
|
|
|
|
|
|
|
|
|
|
|
|
-- | TODO: Make this as thorough as the one for MSSQL/PostgreSQL
|
|
|
|
comparisonExps' ::
|
|
|
|
forall m n. (BackendSchema 'MySQL, MonadSchema n m, MonadError QErr m) =>
|
|
|
|
ColumnType 'MySQL ->
|
|
|
|
m (Parser 'Input n [ComparisonExp 'MySQL])
|
|
|
|
comparisonExps' = P.memoize 'comparisonExps $ \columnType -> do
|
|
|
|
-- see Note [Columns in comparison expression are never nullable]
|
|
|
|
typedParser <- columnParser columnType (G.Nullability False)
|
|
|
|
nullableTextParser <- columnParser (ColumnScalar @'MySQL MySQL.VarChar) (G.Nullability True)
|
|
|
|
textParser <- columnParser (ColumnScalar @'MySQL MySQL.VarChar) (G.Nullability False)
|
|
|
|
let name = P.getName typedParser <> $$(G.litName "_MySQL_comparison_exp")
|
|
|
|
desc = G.Description $ "Boolean expression to compare columns of type "
|
|
|
|
<> P.getName typedParser
|
|
|
|
<<> ". All fields are combined with logical 'AND'."
|
server: remove remnants of query plan caching (fix #1795)
Query plan caching was introduced by - I believe - hasura/graphql-engine#1934 in order to reduce the query response latency. During the development of PDV in hasura/graphql-engine#4111, it was found out that the new architecture (for which query plan caching wasn't implemented) performed comparably to the pre-PDV architecture with caching. Hence, it was decided to leave query plan caching until some day in the future when it was deemed necessary.
Well, we're in the future now, and there still isn't a convincing argument for query plan caching. So the time has come to remove some references to query plan caching from the codebase. For the most part, any code being removed would probably not be very well suited to the post-PDV architecture of query execution, so arguably not much is lost.
Apart from simplifying the code, this PR will contribute towards making the GraphQL schema generation more modular, testable, and easier to profile. I'd like to eventually work towards a situation in which it's easy to generate a GraphQL schema parser *in isolation*, without being connected to a database, and then parse a GraphQL query *in isolation*, without even listening any HTTP port. It is important that both of these operations can be examined in detail, and in isolation, since they are two major performance bottlenecks, as well as phases where many important upcoming features hook into.
Implementation
The following have been removed:
- The entirety of `server/src-lib/Hasura/GraphQL/Execute/Plan.hs`
- The core phases of query parsing and execution no longer have any references to query plan caching. Note that this is not to be confused with query *response* caching, which is not affected by this PR. This includes removal of the types:
- - `Opaque`, which is replaced by a tuple. Note that the old implementation was broken and did not adequately hide the constructors.
- - `QueryReusability` (and the `markNotReusable` method). Notably, the implementation of the `ParseT` monad now consists of two, rather than three, monad transformers.
- Cache-related tests (in `server/src-test/Hasura/CacheBoundedSpec.hs`) have been removed .
- References to query plan caching in the documentation.
- The `planCacheOptions` in the `TenantConfig` type class was removed. However, during parsing, unrecognized fields in the YAML config get ignored, so this does not cause a breaking change. (Confirmed manually, as well as in consultation with @sordina.)
- The metrics no longer send cache hit/miss messages.
There are a few places in which one can still find references to query plan caching:
- We still accept the `--query-plan-cache-size` command-line option for backwards compatibility. The `HASURA_QUERY_PLAN_CACHE_SIZE` environment variable is not read.
https://github.com/hasura/graphql-engine-mono/pull/1815
GitOrigin-RevId: 17d92b254ec093c62a7dfeec478658ede0813eb7
2021-07-27 14:51:52 +03:00
|
|
|
textListParser = fmap openValueOrigin <$> P.list textParser
|
|
|
|
columnListParser = fmap openValueOrigin <$> P.list typedParser
|
2021-07-15 15:44:26 +03:00
|
|
|
pure $ P.object name (Just desc) $ catMaybes <$> sequenceA
|
|
|
|
[ P.fieldOptional $$(G.litName "_is_null") Nothing (bool ANISNOTNULL ANISNULL <$> P.boolean)
|
|
|
|
, P.fieldOptional $$(G.litName "_eq") Nothing (AEQ True . mkParameter <$> typedParser)
|
|
|
|
, P.fieldOptional $$(G.litName "_neq") Nothing (ANE True . mkParameter <$> typedParser)
|
|
|
|
, P.fieldOptional $$(G.litName "_gt") Nothing (AGT . mkParameter <$> typedParser)
|
|
|
|
, P.fieldOptional $$(G.litName "_lt") Nothing (ALT . mkParameter <$> typedParser)
|
|
|
|
, P.fieldOptional $$(G.litName "_gte") Nothing (AGTE . mkParameter <$> typedParser)
|
|
|
|
, P.fieldOptional $$(G.litName "_lte") Nothing (ALTE . mkParameter <$> typedParser)
|
|
|
|
]
|
|
|
|
|
|
|
|
|
|
|
|
offsetParser' :: MonadParse n => Parser 'Both n (SQLExpression 'MySQL)
|
|
|
|
offsetParser' =
|
|
|
|
MySQL.ValueExpression . MySQL.BigValue . fromIntegral <$> P.int
|
|
|
|
|
|
|
|
-- | Various update operators
|
|
|
|
updateOperators' ::
|
|
|
|
Applicative m =>
|
|
|
|
-- | qualified name of the table
|
|
|
|
TableInfo 'MySQL ->
|
|
|
|
-- | update permissions of the table
|
|
|
|
UpdPermInfo 'MySQL ->
|
|
|
|
m (Maybe (InputFieldsParser n [(RQL.Column 'MySQL, IR.UpdOpExpG (UnpreparedValue 'MySQL))]))
|
|
|
|
updateOperators' _table _updatePermissions = pure Nothing
|