Repository navigation
fix(syntax): give numeric Snowflake operands a wrong-kind case and verify 74 more forms - #222
Merged
Merged
Conversation
…d routine modifiers in the tree Parse the Snowflake forms whose recorded verdicts the parser missed: SHOW containers and lone kind words, new SHOW kinds (VOLUMES, DELEGATED AUTHORIZATIONS), DESC and DROP of lone kinds, structured ALTER OPENFLOW, DEPLOYMENT and ROLE, CREATE TASK without properties, table stages after a dot in PUT, and refusals for EXECUTE of a non-kind, CAST to PRECISION, quoted session variables, SET without a variable name and CREATE INDEX on a reserved word. Keep ROOT ONLY, OR ALTER, SECURE, EXTERNAL and TEMPORARY as fields instead of folding them into the kind or dropping them.
…lake docs already allow Remove 27 wrong-outcome obligations whose case either relies on a documented remainder or cannot express an integration type list, each with a docs reason in limits.toml.
…OT POLICY before SET as a snapshot name A single SET or UNSET CONTACT on a table kept the shared ALTER TABLE tree with a raw action; it now keeps the governance tree with the purpose and the contact name. ALTER SNAPSHOT POLICY SET COMMENT names a snapshot called POLICY, as Snowflake reads it, and the nameless SNAPSHOT POLICY SET TAG and UNSET forms are refused.
…outine maps, Iceberg versions and partial CREATE kinds SHOW AVAILABLE LISTINGS keeps its IS_IMPORTED, IS_ORGANIZATION and IS_SHARED_WITH_ME filters. External function HEADERS take bare words and COMPRESSION and the translators take strings. AT reads Iceberg VERSION_TAG and VERSION_REF. A volatile file format keeps VOLATILE. A named argument refuses an unquoted stage, CREATE refuses METRIC alone and an IF after the kind that does not start IF NOT EXISTS.
…wflake forms Each form was read against its documentation page, obligations, cases and parse trees by a reviewer other than the parser's author, and is bound to its fingerprint and definition. 36 reviewed forms were rejected and stay unverified.
Operands the pages name <rows>, <integer>, <*_num>, <*_seconds>, <length>, <precision> or <scale> were read as plain operands, so they had no string literal negative while every name and string operand had one. They now take the integer class: a positive with an integer and a negative with a string literal in its place. The recorded docs/engine conflicts on SKIP_HEADER, MAX_STALENESS and SHOW ... LIMIT keep their suppressions under the new class note.
BY USER, BY and TO SECURITY INTEGRATION were folded into the object kind and the name went to the FOR target. The kind is now DELEGATED AUTHORIZATIONS, the clause is the scope kind and the user or integration is the scope, with its quoting.
…s for the batch 23 receipts Fresh reviewers re-read SHOW RUN METRICS and PARAMETERS IN EXPERIMENT against the wrong-kind criterion after their numeric operands gained obligations. Marks and arity follow the cases regenerated with the new obligations and the batch 23 receipts; the evidence pin follows.
…iven a string Snowflake refuses a string as SEQUENCE_NUMBER, as a data metric function's GROUP LIMIT and as a SHOW row count, so the parser now does too. It parses a string as a routine's MAX_BATCH_ROWS, MAX_BATCH_RETRIES or BATCH_TIMEOUT_SECS, and a semantic view's MAX_STALENESS, which the parser now reads and keeps as a property. ALTER SNAPSHOT POLICY SET reads POLICY as a snapshot name only before COMMENT, the one recorded case; any other SET stays a snapshot policy with its name left out. Tests now also check the scope name of a lone BY USER, the kind of SHOW TASKS with ROOT ONLY, and the routine modifiers kept as fields.
…iven a string Pin the evidence with the receipts for the wrong-kind cases of numeric operands, the remaining SHOW, ALTER and CREATE shapes and the DCM definitions, and refresh the marks they settle. Give the two new SHOW MULTI PARTY APPROVAL POLICIES LIMIT obligations fresh ids instead of the two that the suppressed IN negatives retired, so ids keep their meaning, and point three unobservable records at the integer obligations that replaced theirs.
… MAX_STALENESS as a string Snowflake refuses CREATE INTERACTIVE WAREHOUSE with no name, with IF EXISTS, or with a bare parenthesised list after the name, so the opaque check now knows the kind and refuses them too. A semantic view's MAX_STALENESS takes a string: Snowflake refuses a number there and parses a string, so the parser does the same.
…owflake forms Fresh reviewers read 146 review-only forms in full and approved 107. Four approvals are withdrawn because another reviewer rejected the same gap elsewhere: a missing case for a required function argument, and a repeated list item with no missing case. The other rejections stand until their gaps are fixed. Re-record the baseline. Against the previous one, 74 forms are gained and 53 lost: 41 previously verified forms failed review on gaps the current criteria catch, and 12 forms have a new numeric case that the capture guard holds or whose receipt the importer refuses. Remove the 74 unobservable records whose cases now have receipts and the disagreements the parser now agrees with.
Contributor
Author
Dropped and loosened obligations, re-baseline drops and review details1. 27 security_admin derived obligations suppressed with docs reasons (0b426da)Each was a derived rejection whose SQL the docs also accept, or a class that does not apply. Case changes: 19 evidence cases removed with them.
2. Numeric operand class (4fc6564)
3. Other changes found by the independent review (none loosens an obligation)
4. Re-baseline drops not recovered by review (12 forms)
5. Previously verified forms that failed fresh review (41)
6. Reviewer conflicts
7. Spot-checked approvals
Each grammar matches its docs Syntax block, every case agrees with its receipt, and the new 8. Reachable forms after this change (621 catalogued forms unverified)
|
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
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.
Why
Numeric Snowflake operands had no wrong-kind case, so reviewers rejected forms that otherwise held, and some recorded verdicts disagreed with the parser.
Changes
<integer>,<rows>,<num>and similar) get the integer class and a "string in place of a number" case. Name operands are unchanged.Verification
make check-cipasses on the final head.