The names for the SSL mode options were incorrect, added test for making
sure the built url can be parsed to ensure internal consistency at
least. Also changed "ssl-mode" to "sslmode" as it seems more standard
and sqlx can parse both.
* feat: Add set_update_hook on SqliteConnection
* refactor: Address PR comments
* fix: Expose UpdateHookResult for public use
---------
Co-authored-by: John Smith <asserta4@gmail.com>
* Revert "fix(ci): disable `mariadb_verylatest` pass for RusTLS"
This reverts commit c78425b3ac.
MariaDB 11.4.2 now includes the upstream fix for this.
Closes: 3091
* ci: mariadb 11.4 now GA/Stable
* feat: support `NonZero*` scalar types
This commits adds `Type`, `Encode`, and `Decode` impls for all the
`NonZero*` types from the standard library. They are implemented as
direct proxies to their primitive counterparts, except that when
decoding, the values are checked to not be zero.
* fixup!: remove `non-zero` cargo feature
* fixup!: make `non-zero` module private
* fixup!: rebase and fix trait impls
* convert logger to output a query graph
* avoid duplicating branch paths to shrink output graph
* separate different branching paths
* include all branches which found unique states
* track the reason for ending each branches execution
* track the result type of each branch
* make edges rely on history index instead of program_id, to avoid errors when looping
* add state diff to query graph
* drop redundant table info
* rework graph to show state changes, rework logger to store state snapshots
* show state on the previous operation
* gather duplicate state changes into clusters to reduce repetition
* draw invisible connections between unknown instructions by program_i
* clean up dot format string escaping
* add test case from #1960 (update returning all columns)
* add tests for #2939 (update returning only the PK column)
* allow inserting into a table using only the index
* improve null handling of IfNull, fix output type of NewRowId
* add NoResult nodes for branches which don't log a result, as a sanity check
* add short-circuit to all logging operations
* remove duplicate logging checks, and make logging enabled/disabled consistently depend on sqlx::explain instead of sqlx for capture & sqlx::explain for output
* add failing test for awkwardly nested/filtered count subquery
* handle special case of return operation to fix failing test
* require trace log level instead of using whatever log level statement logging was configured to use
* Make encode and encode_by_ref fallible
This only changes the trait for now and makes it compile, calling .expect() on all users. Those will be removed in a later commit.
* PgNumeric: Turn TryFrom Decimal to an infallible From
* Turn panics in Encode implementations into errors
* Add Encode error analogous to the Decode error
* Propagate decode errors through Arguments::add
This pushes the panics one level further to mostly bind calls. Those will also be removed later.
* Only check argument encoding at the end
* Use Result in Query internally
* Implement query_with functions in terms of _with_result
* Surface encode errors when executing a query.
* Remove remaining panics in AnyConnectionBackend implementations
* PostgreSQL BigDecimal: Return encode error immediately
* Arguments: Add len method to report how many arguments were added
* Query::bind: Report which argument failed to encode
* IsNull: Add is_null method
* MySqlArguments: Replace manual bitmap code with NullBitMap helper type
* Roll back buffer in MySqlArguments if encoding fails
* Roll back buffer in SqliteArguments if encoding fails
* Roll back PgArgumentBuffer if encoding fails
* fix: make resolve_blocking not take ownership of path
When using sqlx_macros_unstable the codepath taken further uses the path
variable and it is more convenient to not take ownership but instead
pass references to needed functions.
* fix: change &PathBuf to &Path in resolve_blocking
* fix: sqlite pragma order for auto_vacuum
Setting the auto_vacuum pragma must come before setting the journal
mode otherwise it won't apply.
* fix: better documentation for auto_vacuum
Co-authored-by: Austin Bonander <austin.bonander@gmail.com>
---------
Co-authored-by: Austin Bonander <austin.bonander@gmail.com>
When using the 'Any' driver with MySQL backend, fetch_optional
does not return the connection to the pool if no results
are returned from the query. This is due to not all of the packets
being read from the underlying stream.
This fix continues to read result packets from the stream until they
have all been exhausted (just like the normal MySql drivers
implementation of fetch_optional). In general, a better refactoring would
be to call the MySQL fetch_optional code in the Any driver, rather than
re-implementing and duplicating code.
* Include test case for regular subtransactions
While using COPY and subtransactions I kept running into errors.
This test case documents that error, it currently fails with:
Error: encountered unexpected or invalid data: expecting ParseComplete but received CommandComplete
* PostgreSQL Copy: Consume ReadyForQuery on error
When a COPY statement was in error inside a subtransaction,
a Protocol Error used to be raised. By consuming the ReadyForQuery
message when there is an error, we no longer have this issue.
* fix(postgres) : int type conversion while decoding
* Check value buffer len when decoding integer type in postgres
* decode from octal for postgres i8
---------
Co-authored-by: RaghavRox <66472843+RaghavRox@users.noreply.github.com>
* test: add a failing test
* feat: add no_tx to migration struct
* feat: execute migration with no tx block
* fix: expected string literal compilation error
* test: update no tx to content comment
* refactor: use the sql comment instead of file name semantics
* docs: remove no_tx from file format comment
* fix: remove filename matches
* fix: messed up merge
* refactor: dedupe migration
* fix: move comment to where it makes sense
* fix: linter error
Breaking changes:
* integer decoding will now loudly error on overflow instead of silently truncating.
* some usages of the query!() macros might change an i32 to an i64.
Also adds support for *decoding* `u64`s because there's no reason not to.
Closes#3179
