Deprecate mixing qualified and unqualified names #6677
Merged
Conversation
This file contains hidden or 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
morozov
commented
Dec 28, 2024
| @@ -228,7 +228,10 @@ public function testHasTableForQuotedAsset(): void | |||
|
|
|||
| public function testHasNamespace(): void | |||
| { | |||
| $schema = new Schema(); | |||
| $schemaConfig = new SchemaConfig(); | |||
| $schemaConfig->setName('public'); | |||
There was a problem hiding this comment.
This and other similarly modified tests operate unqualified and qualified names. From now on, this is only allowed if the schema has a default namespace configuration.
greg0ire
reviewed
Dec 29, 2024
| * Indicates whether the schema uses unqualified names for its objects. Once this flag is set to true, it won't be | ||
| * unset even after the objects with unqualified names have been dropped from the schema. | ||
| */ | ||
| private bool $usesUnqualifiedNames = false; |
There was a problem hiding this comment.
Is this flag write once? Should it be marked as such by using @readonly?
There was a problem hiding this comment.
It's not exactly write once, it can be flipped once from false to true. @readonly might be confusing, if the reader assumes the same contract as the one provided by the readonly modifier:
[...] which prevents modification of the property after initialization.
This flag is initialized as f
8000
alse and can be modified at any point in time to true or never modified.
greg0ire
approved these changes
Dec 29, 2024
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Outdated suggestions cannot be applied.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.
Technically, each of the three commits could be contributed as a separate pull request but I'll publish them altogether in the interest of time as each of them depends on the previous one.
1. Population of the current SQL Server schema
Technically, the fact that the current SQL Server schema isn't populated during introspection is a bug. Once introspected, a PostgreSQL schema allows referencing its objects by unqualified or qualified name but the SQL Server schema doesn't. This commit addresses this issue and reworks an existing test to cover this scenario.
2. Deprecation of mixing qualified and unqualified names in a schema
Currently, a schema allows registering objects with a mixture of unqualified names (e.g.
customers) and qualified names (e.g.public.customers). It compares names as strings, so in the absence of a default schema name, "customers" ≠ "public.customers", however, it's a logical error because these two values are incomparable (again, similar to "5" and "5 meters").In order to make the schema API more robust, mixing incomparable types of names is deprecated.
Using unqualified names
Unqualified names can be used to add objects to or reference objects in a schema only if doesn't contain objects with qualified names or has a default namespace configuration.
Using qualified names
Qualified names can be used to add objects to or reference objects in a schema only if doesn't contain objects with unqualified names.
Considered alternatives
Whether a given schema uses unqualified or qualified names is determined at runtime. An even more robust approach would be parameterize the schema class with the type of its objects' names and use this information for static analysis. However, the type of the names is defined by the target database platform, which is determined at runtime based on connection parameters, so the static typing is not feasible.
3. Deprecation of
AbstractAsset::isIdentifierQuoted()Once these changes are merged into 5.0.x, it will be possible to stop using
AbstractAsset::isIdentifierQuoted(). At that point, we should be able to remove it.The problem with this method is that it accepts a name that potentially consists of multiple identifiers and returns the value based on whether the first one is quoted. This is incorrect as each identifier may or may not be quoted individually.