Experiment with multi-file schema handling in PSL #4243
Merged
+967
−554
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
CodSpeed Performance ReportMerging #4243 will degrade performances by 5.45%Comparing Summary
Benchmarks breakdown
|
tomhoule
commented
Sep 15, 2023
tomhoule
commented
Sep 15, 2023
tomhoule
commented
Sep 15, 2023
tomhoule
commented
Sep 15, 2023
tomhoule
force-pushed
the
psl-multi-file-schema
branch
from
14c28e3
to
ccf1968
Compare
janpio
changed the title
SevInf
force-pushed
the
psl-multi-file-schema
branch
from
ccf1968
to
f34efb7
Compare
WASM Query Engine file Size
|
❌ WASM query-engine performance will worsen by 1.53%Full benchmark reportAfter changes in 3fee3aa |
This commit implements multi-file schema handling in the Prisma Schema Language.
At a high level, instead of accepting a single string, `psl::validate_multi_file()` is an alternative to `psl::validate()` that accepts something morally equivalent to:
```json
{
"./prisma/schema/a.prisma": "datasource db { ... }",
"./prisma/schema/nested/b.prisma": "model Test { ... }"
}
```
There are tests for PSL validation with multiple schema files, but most of the rest of engines still consumes the single file version of `psl::validate()`. The implementation and the return type are shared between `psl::validate_multi_file()` and `psl::validate()`, so the change is completely transparent, other than the expectation of passing in a list of (file_name, file_contents) instead of a single string. The `psl::validate()` entry point should behave exactly the same as `psl::multi_schema()` with a single file named `schema.prisma`. In particular, it has the exact same return type.
Implementation
==============
This is achieved by extending `Span` to contain, in addition to a start and end offset, a `FileId`. The `FileId` is a unique identifier for a file and its parsed `SchemaAst` inside `ParserDatabase`. The identifier types for AST items in `ParserDatabase` are also extended to contain the `FileId`, so that they can be uniquely referred to in the context of the (multi-file) schema. After the analysis phase (the `parser_database` crate), consumers of the analyzed schema become multi-file aware completely transparently, no change is necessary in the other engines.
The only changes that will be required at scattered points across the codebase are the `psl::validate()` call sites that will need to receive a `Vec<Box<Path>, SourceFile>` instead of a single `SourceFile`. This PR does _not_ deal with that, but it makes where these call sites are obvious by what entry points they use: `psl::validate()`, `psl::parse_schema()` and the various `*_assert_single()` methods on `ParserDatabase`.
The PR contains tests confirming that schema analysis, validation and displaying diagnostics across multiple files works as expected.
Status of this PR
=================
This is going to be directly mergeable after review, and it will not affect the current schema handling behaviour when dealing with a single schema file.
Next steps
==========
- Replace all calls to `psl::validate()` with calls to `psl::validate_multi_file()`.
- The `*_assert_single()` calls should be progressively replaced with their multi-file counterparts across engines.
- The language server should start sending multiple files to prisma-schema-wasm in all calls. This is not in the spirit of the language server spec, but that is the most immediate solution. We'll have to make `range_to_span()` in `prisma-fmt` multi-schema aware by taking a FileId param.
Links
=====
Relevant issue: prisma/prisma#2377
Also see the [internal design doc](https://www.notion.so/prismaio/Multi-file-Schema-24d68fe8664048ad86252fe446caac24?d=68ef128f25974e619671a9855f65f44d#2889a038e68c4fe1ac9afe3cd34978bd).
SevInf
force-pushed
the
psl-multi-file-schema
branch
from
f34efb7
to
1e42d35
Compare
jkomyno
reviewed
Apr 4, 2024
jkomyno
reviewed
Apr 4, 2024
jkomyno
reviewed
Apr 4, 2024
jkomyno
reviewed
Apr 4, 2024
This comment was marked as duplicate.
This comment was marked as duplicate.
Sorry, something went wrong.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This commit implements multi-file schema handling in the Prisma Schema Language.
At a high level, instead of accepting a single string,
psl::validate_multi_file()is an alternative topsl::validate()that accepts something morally equivalent to:{ "./prisma/schema/a.prisma": "datasource db { ... }", "./prisma/schema/nested/b.prisma": "model Test { ... }" }There are tests for PSL validation with multiple schema files, but most of the rest of engines still consumes the single file version of
psl::validate(). The implementation and the return type are shared betweenpsl::validate_multi_file()andpsl::validate(), so the change is completely transparent, other than the expectation of passing in a list of (file_name, file_contents) instead of a single string. Thepsl::validate()entry point should behave exactly the same aspsl::multi_schema()with a single file namedschema.prisma. In particular, it has the exact same return type.Implementation
This is achieved by extending
Spanto contain, in addition to a start and end offset, aFileId. TheFileIdis a unique identifier for a file and its parsedSchemaAstinsideParserDatabase. The identifier types for AST items inParserDatabaseare also extended to contain theFileId, so that they can be uniquely referred to in the context of the (multi-file) schema. After the analysis phase (theparser_databasecrate), consumers of the analyzed schema become multi-file aware completely transparently, no change is necessary in the other engines.The only changes that will be required at scattered points across the codebase are the
psl::validate()call sites that will need to receive aVec<Box<Path>, SourceFile>instead of a singleSourceFile. This PR does not deal with that, but it makes where these call sites are obvious by what entry points they use:psl::validate(),psl::parse_schema()and the various*_assert_single()methods onParserDatabase.The PR contains tests confirming that schema analysis, validation and displaying diagnostics across multiple files works as expected.
Status of this PR
This is going to be directly mergeable after review, and it will not affect the current schema handling behaviour when dealing with a single schema file.
Next steps
psl::validate()with calls topsl::validate_multi_file().*_assert_single()calls should be progressively replaced with their multi-file counterparts across engines.range_to_span()inprisma-fmtmulti-schema aware by taking a FileId param.Links
Relevant issue: prisma/prisma#2377
Also see the internal design doc.
Close prisma/team-orm#1034