fix: various mssql compat fixes - #5483
Conversation
CodSpeed Performance ReportMerging #5483 will not alter performanceComparing Summary
|
WASM Query Engine file Size
|
6f6996b to
1a4c112
Compare
1a4c112 to
8ccfe3b
Compare
| new::interactive_tx::interactive_tx::batch_queries_failure | ||
| new::regressions::max_integer::max_integer::unfitted_int_should_fail | ||
| new::regressions::max_integer::max_integer::unfitted_int_should_fail_mssql | ||
| new::regressions::prisma_7434::not_in_chunking::not_in_batch_filter | ||
| queries::aggregation::group_by::aggregation_group_by::group_by_ordering_sum_aggregation | ||
| queries::chunking::chunking::order_by_aggregation_should_fail | ||
| queries::filters::self_relation_regression::sr_regression::all_categories | ||
| raw::sql::typed_output::typed_output::unknown_type_mssql | ||
| writes::data_types::bytes::bytes::using_bytes_field | ||
| writes::data_types::decimal::decimal::using_decimal_as_id | ||
| writes::data_types::native_types::sql_server::sql_server::native_decimal_type | ||
| writes::nested_mutations::already_converted::nested_connect_inside_update::connect_inside_update::p1_c1req_rel_child_idempotent |
There was a problem hiding this comment.
What do we plan to do with these failures?
There was a problem hiding this comment.
@jkomyno most of these failures affect all databases in the same way so they'll be tackled when we fix them for other databases, I think only 4 failures are specific to MSSQL:
- raw::sql::typed_output::typed_output::unknown_type_mssql - we raise a slightly different error that we're supposed to, we will likely need a new custom error variant
- writes::data_types::bytes::bytes::using_bytes_field - the test works fine until it hits
SET field = nullon a bytes field, I've tried multiple workarounds but it keeps failing, will need to be investigated more - writes::data_types::decimal::decimal::using_decimal_as_id - it works fine except at the end of the test the values gets read as 1000000000.0000000000000001 instead of 1000000000
- native_decimal_type - minor different in number formatting
| // TODO: investigate why the adapter raises a different error | ||
| CONFIG.connector_version() == Some("mssql.js.wasm") |
There was a problem hiding this comment.
Can you please create a new task for this?
There was a problem hiding this comment.
Created 2 tickets for remaining mssql failures
https://linear.app/prisma-company/issue/ORM-1090/fix-failing-data-type-tests
https://linear.app/prisma-company/issue/ORM-1091/fix-inconsistent-concurrentdeadlocked-transaction-behavior
There was a problem hiding this comment.
for raw::sql::typed_output::typed_output::unknown_type_mssql, I think we should just cover it under a non-db specific ticket where go through all error-related tests and make sure all error messages are aligned (the main difficulty is that DA QE and DA QC currently use the same snapshots and it's difficult to align them perfectly - the error messages get formatted properly, they just don't always match 1:1)
2 participants
ORM-1056
Starts running tests against MSSQL with the query compiler.
I've had to refactor some code to utilize
RowSink, because there were many edge case plans with missinguniquenodes that were causing problems./prisma-branch fix/misc-mssql-fixes