All notable changes to this project will be documented in this file.
The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.
10 pull requests were merged this release cycle.
This release addresses a few regressions that have occurred, and refines SQLx's MSRV policy (see the FAQ).
- [#3424]: Remove deprecated feature-names from
Cargo.toml
files in examples [@carschandler]
- [#3403]: Fix (#3395) sqlx::test macro in 0.8 [@joeydewaal]
- [#3411]: fix: Use rfc3339 to decode date from text [@pierre-wehbe]
- [#3453]: fix(#3445): PgHasArrayType [@joeydewaal]
- Fixes
#[sqlx(no_pg_array)]
being forbidden on#[derive(Type)]
structs.
- Fixes
- [#3454]: fix: non snake case warning [@joeydewaal]
- [#3459]: Pgsql cube type compile fail [@kdesjard]
- [#3465]: fix(postgres): max number of binds is 65535, not 32767 (regression) [@abonander]
- [#3467]: fix cancellation issues with
PgListener
,PgStream::recv()
[@abonander]- Fixes cryptic
unknown message: "\\0"
error
- Fixes cryptic
- [#3474]: Fix try_get example in README.md [@luveti]
16 pull requests were merged this release cycle.
This release contains a fix for RUSTSEC-2024-0363.
Postgres users are advised to upgrade ASAP as a possible exploit has been demonstrated: launchbadge#3440 (comment)
MySQL and SQLite do not appear to be exploitable, but upgrading is recommended nonetheless.
- [#3421]: correct spelling of
MySqlConnectOptions::no_engine_substitution()
[@kolinfluence]- Deprecates
MySqlConnectOptions::no_engine_subsitution()
(oops) in favor of the correctly spelled version.
- Deprecates
- [#3376]: doc: hide
spec_error
module [@abonander]- This is a helper module for the macros and was not meant to be exposed.
- It is not expected to receive any breaking changes for the 0.8.x release, but is not designed as a public API. Use at your own risk.
- [#3382]: feat: bumped to
libsqlite3-sys=0.30.1
to support sqlite 3.46 [@CommanderStorm] - [#3385]: chore(examples):Migrated the pg-chat example to ratatui [@CommanderStorm]
- [#3399]: Upgrade to rustls 0.23 [@djc]
- RusTLS now has pluggable cryptography providers:
ring
(the existing implementation), andaws-lc-rs
which has optional FIPS certification. - The existing features activating RusTLS (
runtime-tokio-rustls
,runtime-async-std-rustls
,tls-rustls
) enable thering
provider of RusTLS to match the existing behavior so this should not be a breaking change. - Switch to the
tls-rustls-aws-lc-rs
feature to use theaws-lc-rs
provider.- If using
runtime-tokio-rustls
orruntime-async-std-rustls
, this will necessitate switching to the appropriate non-legacy runtime feature:runtime-tokio
orruntime-async-std
- If using
- See the RusTLS README for more details: https://github.com/rustls/rustls?tab=readme-ov-file#cryptography-providers
- RusTLS now has pluggable cryptography providers:
- [#2786]: fix(sqlx-cli): do not clean sqlx during prepare [@cycraig]
- [#3354]: sqlite: fix inconsistent read-after-write [@ckampfe]
- [#3371]: Fix encoding and decoding of MySQL enums in
sqlx::Type
[@alu] - [#3374]: fix: usage of
node12
inSQLx
action [@hamirmahal] - [#3380]: chore: replace structopt with clap in examples [@tottoto]
- [#3381]: Fix CI after Rust 1.80, remove dead feature references [@abonander]
- [#3384]: chore(tests): fixed deprecation warnings [@CommanderStorm]
- [#3386]: fix(dependencys):bumped cargo_metadata to
v0.18.1
to avoid yankedv0.14.3
[@CommanderStorm] - [#3389]: fix(cli): typo in error for required DB URL [@ods]
- [#3417]: Update version to 0.8 in README [@soucosmo]
- [#3441]: fix: audit protocol handling [@abonander]
- This addresses RUSTSEC-2024-0363 and includes regression tests for MySQL, Postgres and SQLite.
70 pull requests were merged this release cycle.
#2697 was merged the same day as release 0.7.4 and so was missed by the automatic CHANGELOG generation.
- [#2697]: fix(macros): only enable chrono when time is disabled [@saiintbrisson]
- [#2973]: Generic Associated Types in Database, replacing HasValueRef, HasArguments, HasStatement [@nitn3lav]
- [#2482]: chore: bump syn to 2.0 [@saiintbrisson]
- Deprecated type ascription syntax in the query macros was removed.
- [#2736]: Fix describe on PostgreSQL views with rules [@tsing]
- Potentially breaking: nullability inference changes for Postgres.
- [#2869]: Implement PgHasArrayType for all references [@tylerhawkes]
- Conflicts with existing manual implementations.
- [#2940]: fix: Decode and Encode derives (#1031) [@benluelo]
- Changes lifetime obligations for field types.
- [#3064]: Sqlite explain graph [@tyrelr]
- Potentially breaking: nullability inference changes for SQLite.
- [#3123]: Reorder attrs in sqlx::test macro [@bobozaur]
- Potentially breaking: attributes on
#[sqlx::test]
usages are applied in the correct order now.
- Potentially breaking: attributes on
- [#3126]: Make Encode return a result [@FSMaxB]
- [#3130]: Add version information for failed cli migration (#3129) [@FlakM]
- Breaking changes to
MigrateError
.
- Breaking changes to
- [#3181]: feat: no tx migration [@cleverjam]
- (Postgres only) migrations that should not run in a transaction can be flagged by adding
-- no-transaction
to the beginning. - Breaking change: added field to
Migration
- (Postgres only) migrations that should not run in a transaction can be flagged by adding
- [#3184]: [BREAKING} fix(sqlite): always use
i64
as intermediate when decoding [@abonander]- 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.
- [#3252]: fix
#[derive(sqlx::Type)]
in Postgres [@abonander]- Manual implementations of PgHasArrayType for enums will conflict with the generated one. Delete the manual impl or add
#[sqlx(no_pg_array)]
where conflicts occur. - Type equality for PgTypeInfo is now schema-aware.
- Manual implementations of PgHasArrayType for enums will conflict with the generated one. Delete the manual impl or add
- [#3329]: fix: correct handling of arrays of custom types in Postgres [@abonander]
- Potential breaking change:
PgTypeInfo::with_name()
infers types that start with_
to be arrays of the un-prefixed type. Wrap type names in quotes to bypass this behavior.
- Potential breaking change:
- [#3356]: breaking: fix name collision in
FromRow
, returnError::ColumnDecode
forTryFrom
errors [@abonander]- Breaking behavior change: errors with
#[sqlx(try_from = "T")]
now returnError::ColumnDecode
instead ofError::ColumnNotFound
. - Breaking because
#[sqlx(default)]
on an individual field or the struct itself would have previously suppressed the error. This doesn't seem like good behavior as it could result in some potentially very difficult bugs.- Instead, create a wrapper implementing
From
and apply the default explicitly.
- Instead, create a wrapper implementing
- Breaking behavior change: errors with
- [#3337]: allow rename with rename_all (close #2896) [@DirectorX]
- Changes the precedence of
#[sqlx(rename)]
and#[sqlx(rename_all)]
to match the expected behavior (rename
wins).
- Changes the precedence of
- [#3285]: fix: use correct names for sslmode options [@lily-mosquitoes]
- Changes the output of
ConnectOptions::to_url_lossy()
to match what parsing expects.
- Changes the output of
- [#2917]: Add Debug impl for PgRow [@g-bartoszek]
- [#3113]: feat: new derive feature flag [@saiintbrisson]
- [#3154]: feat: add
MySqlTime
, auditmysql::types
for panics [@abonander] - [#3188]: feat(cube): support postgres cube [@jayy-lmao]
- [#3244]: feat: support
NonZero*
scalar types [@AlphaKeks] - [#3260]: feat: Add set_update_hook on SqliteConnection [@gridbox]
- [#3291]: feat: support the Postgres Bool type for the Any driver [@etorreborre]
- [#3293]: Add LICENSE-* files to crates [@LecrisUT]
- [#3303]: add array support for NonZeroI* in postgres [@JohannesIBK]
- [#3311]: Add example on how to use Transaction as Executor [@Lachstec]
- [#3343]: Add support for PostgreSQL HSTORE data type [@KobusEllis]
- [#2652]: MySQL: Remove collation compatibility check for strings [@alu]
- [#2960]: Removed
Send
trait bound from argument binding [@bobozaur] - [#2970]: refactor: lift type mappings into driver crates [@abonander]
- [#3148]: Bump libsqlite3-sys to v0.28 [@NfNitLoop]
- Note: version bumps to
libsqlite3-sys
are not considered breaking changes as per our semver guarantees.
- Note: version bumps to
- [#3265]: perf: box
MySqlConnection
to reduce sizes of futures [@stepantubanov] - [#3352]: chore:added a testcase for
sqlx migrate add ...
[@CommanderStorm] - [#3340]: ci: Add job to check that sqlx builds with its declared minimum dependencies [@iamjpotts]
- [#2702]: Constrain cyclic associated types to themselves [@BadBastion]
- [#2954]: Fix several inter doc links [@ralpha]
- [#3073]: feat(logging): Log slow acquires from connection pool [@iamjpotts]
- [#3137]: SqliteConnectOptions::filename() memory fix (#3136) [@hoxxep]
- [#3138]: PostgreSQL Bugfix: Ensure connection is usable after failed COPY inside a transaction [@feikesteenbergen]
- [#3146]: fix(sqlite): delete unused
ConnectionHandleRaw
type [@abonander] - [#3162]: Drop urlencoding dependency [@paolobarbolini]
- [#3165]: Bump deps that do not need code changes [@GnomedDev]
- [#3167]: fix(ci): use
docker compose
instead ofdocker-compose
[@abonander] - [#3172]: fix: Option decoding in any driver [@pxp9]
- [#3173]: fix(postgres) : int type conversion while decoding [@RaghavRox]
- [#3190]: Update time to 0.3.36 [@BlackSoulHub]
- [#3191]: Fix unclean TLS shutdown [@levkk]
- [#3194]: Fix leaking connections in fetch_optional (#2647) [@danjpgriffin]
- [#3216]: security: bump rustls to 0.21.11 [@toxeus]
- [#3230]: fix: sqlite pragma order for auto_vacuum [@jasonish]
- [#3233]: fix: get_filename should not consume self [@jasonish]
- [#3234]: fix(ci): pin Rust version, ditch unmaintained actions [@abonander]
- [#3236]: fix: resolve
path
ownership problems when usingsqlx_macros_unstable
[@lily-mosquitoes] - [#3254]: fix: hide
sqlx_postgres::any
[@Zarathustra2] - [#3266]: ci: MariaDB - add back 11.4 and add 11.5 [@grooverdan]
- [#3267]: ci: syntax fix [@grooverdan]
- [#3271]: docs(sqlite): fix typo - unixtime() -> unixepoch() [@joelkoen]
- [#3276]: Invert boolean for
migrate
error message. (#3275) [@nk9] - [#3279]: fix Clippy errors [@abonander]
- [#3288]: fix: sqlite update_hook char types [@jasonish]
- [#3297]: Pass the
persistent
query setting when preparing queries with theAny
driver [@etorreborre] - [#3298]: Track null arguments in order to provide the appropriate type when converting them. [@etorreborre]
- [#3312]: doc: Minor rust docs fixes [@SrGesus]
- [#3327]: chore: fixed one usage of
select_input_type!()
being unhygenic [@CommanderStorm] - [#3328]: fix(ci): comment not separated from other characters [@hamirmahal]
- [#3341]: refactor: Resolve cargo check warnings in postgres examples [@iamjpotts]
- [#3346]: fix(postgres): don't panic if
M
orC
Notice fields are not UTF-8 [@YgorSouza] - [#3350]: fix:the
json
-feature should activatesqlx-postgres?/json
as well [@CommanderStorm] - [#3353]: fix: build script new line at eof [@Zarthus]
- (no PR): activate
clock
andstd
features ofworkspace.dependencies.chrono
.
38 pull requests were merged this release cycle.
This is officially the last release of the 0.7.x release cycle.
As of this release, development of 0.8.0 has begun on main
and only high-priority bugfixes may be backported.
- [#2891]: feat: expose getters for connect options fields [@saiintbrisson]
- [#2902]: feat: add
to_url_lossy
to connect options [@lily-mosquitoes] - [#2927]: Support
query!
for cargo-free systems [@kshramt] - [#2997]: doc(FAQ): add entry explaining prepared statements [@abonander]
- [#3001]: Update README to clarify MariaDB support [@iangilfillan]
- [#3004]: feat(logging): Add numeric elapsed time field elapsed_secs [@iamjpotts]
- [#3007]: feat: add
raw_sql
API [@abonander]- This hopefully makes it easier to find how to execute statements which are not supported by the default
prepared statement interfaces
query*()
andquery!()
. - Improved documentation across the board for the
query*()
functions. - Deprecated:
execute_many()
andfetch_many()
on interfaces that use prepared statements.- Multiple SQL statements in one query string were only supported by SQLite because its prepared statement interface is the only way to execute SQL. All other database flavors forbid multiple statements in one prepared statement string as an extra defense against SQL injection.
- The new
raw_sql
API retains this functionality because it explicitly does not use prepared statements. Raw or text-mode query interfaces generally allow multiple statements in one query string, and this is supported by all current databases. Due to their nature, however, one cannot use bind parameters with them. - If this change affects you, an issue is open for discussion: launchbadge#3108
- This hopefully makes it easier to find how to execute statements which are not supported by the default
prepared statement interfaces
- [#3011]: Added support to IpAddr with MySQL/MariaDB. [@Icerath]
- [#3013]: Add default implementation for PgInterval [@pawurb]
- [#3018]: Add default implementation for PgMoney [@pawurb]
- [#3026]: Update docs to reflect support for MariaDB data types [@iangilfillan]
- [#3037]: feat(mysql): allow to connect with mysql driver without default behavor [@darkecho731]
- [#2900]: Show latest url to docs for macro.migrate [@Vrajs16]
- [#2914]: Use
create_new
instead ofatomic-file-write
[@mattfbacon] - [#2926]: docs: update example for
PgConnectOptions
[@Fyko] - [#2989]: sqlx-core: Remove dotenvy dependency [@joshtriplett]
- [#2996]: chore: Update ahash to 0.8.7 [@takenoko-gohan]
- [#3006]: chore(deps): Replace unmaintained tempdir crate with tempfile [@iamjpotts]
- [#3008]: chore: Ignore .sqlx folder created by running ci steps locally [@iamjpotts]
- [#3009]: chore(dev-deps): Upgrade env_logger from 0.9 to 0.11 [@iamjpotts]
- [#3010]: chore(deps): Upgrade criterion to 0.5.1 [@iamjpotts]
- [#3050]: Optimize SASL auth in sqlx-postgres [@mirek26]
- [#3055]: Set TCP_NODELAY option on TCP sockets [@mirek26]
- [#3065]: Improve max_lifetime handling [@mirek26]
- [#3072]: Change the name of "inner" function generated by
#[sqlx::test]
[@ciffelia] - [#3083]: Remove sha1 because it's not being used in postgres [@rafaelGuerreiro]
- [#2898]: Fixed docs [@Vrajs16]
- [#2905]: fix(mysql): Close prepared statement if persistence is disabled [@larsschumacher]
- [#2913]: Fix handling of deferred constraints [@Thomasdezeeuw]
- [#2919]: fix duplicate "`" in FromRow "default" attribute doc comment [@shengsheng]
- [#2932]: fix(postgres): avoid unnecessary flush in PgCopyIn::read_from [@tsing]
- [#2955]: Minor fixes [@Dawsoncodes]
- [#2963]: Fixed ReadMe badge styling [@tadghh]
- [#2976]: fix: AnyRow not support PgType::Varchar [@holicc]
- [#3053]: fix: do not panic when binding a large BigDecimal [@Ekleog]
- [#3056]: fix: spans in sqlite tracing (#2876) [@zoomiti]
- [#3089]: fix(migrate): improve error message when parsing version from filename [@abonander]
- [#3098]: Migrations fixes [@abonander]
- Unhides
sqlx::migrate::Migrator
. - Improves I/O error message when failing to read a file in
migrate!()
.
- Unhides
38 pull requests were merged this release cycle.
- [#2478]: feat(citext): support postgres citext [@hgranthorner]
- [#2545]: Add
fixtures_path
in sqlx::test args [@ripa1995] - [#2665]: feat(mysql): support packet splitting [@tk2217]
- [#2752]: Enhancement #2747 Provide
fn PgConnectOptions::get_host(&self)
[@boris-lok] - [#2769]: Customize the macro error message based on the metadata [@Nemo157]
- [#2793]: derived Hash trait for PgInterval [@yasamoka]
- [#2801]: derive FromRow: sqlx(default) for all fields [@grgi]
- [#2827]: Add impl
FromRow
for the unit type [@nanoqsh] - [#2871]: Add
MySqlConnectOptions::get_database()
[@shiftrightonce] - [#2873]: Sqlx Cli: Added force flag to drop database for postgres [@Vrajs16]
- [#2894]: feat:
Text
adapter [@abonander]
- [#2701]: Remove documentation on offline feature [@Baptistemontan]
- [#2713]: Add additional info regarding using Transaction and PoolConnection as… [@satwanjyu]
- [#2770]: Update README.md [@snspinn]
- [#2797]: doc(mysql): document behavior regarding
BOOLEAN
and the query macros [@abonander] - [#2803]: Don't use separate temp dir for query jsons (2) [@mattfbacon]
- [#2819]: postgres begin cancel safe [@conradludgate]
- [#2832]: Update extra_float_digits default to 2 instead of 3 [@brianheineman]
- [#2865]: Update Faq - Bulk upsert with optional fields [@Vrajs16]
- [#2880]: feat: use specific message for slow query logs [@abonander]
- [#2882]: Do not require db url for prepare [@tamasfe]
- [#2890]: doc(sqlite): cover lack of
NUMERIC
support [@abonander] - [No PR]: Upgraded
libsqlite3-sys
to 0.27.0- Note: linkage to
libsqlite3-sys
is considered semver-exempt; see the release notes for 0.7.0 below for details.
- Note: linkage to
- [#2640]: fix: sqlx::macro db cleanup race condition by adding a margin to current timestamp [@fhsgoncalves]
- [#2655]: [fix] Urlencode when passing filenames to sqlite3 [@uttarayan21]
- [#2684]: Make PgListener recover from UnexpectedEof [@hamiltop]
- [#2688]: fix: Make rust_decimal and bigdecimal decoding more lenient [@cameronbraid]
- [#2754]: Is tests/x.py maintained? And I tried fix it. [@qwerty2501]
- [#2784]: fix: decode postgres time without subsecond [@granddaifuku]
- [#2806]: Depend on version of async-std with non-private spawn-blocking [@A248]
- [#2820]: fix: correct decoding of
rust_decimal::Decimal
for high-precision values [@abonander] - [#2822]: issue #2821 Update error handling logic when opening a TCP connection [@anupj]
- [#2826]: chore: bump some sqlx-core dependencies [@djc]
- [#2838]: Fixes rust_decimal scale for Postgres [@jkleinknox]
- [#2847]: Fix comment in
sqlx migrate add
help text [@cryeprecision] - [#2850]: fix(core): avoid unncessary wakeups in
try_stream!()
[@abonander] - [#2856]: Prevent warnings running
cargo build
[@nyurik] - [#2864]: fix(sqlite): use
AtomicUsize
for thread IDs [@abonander] - [#2892]: Fixed force dropping bug [@Vrajs16]
23 pull requests were merged this release cycle.
- [#2121]: Add JSON support to
FromRow
derive [@95ulisse] - [#2533]: Implement mysql_clear_password [@ldanilek]
- [#2538]: cli: add --target-version CLI flags for migrate run/revert [@inahga]
- [#2577]: supplement Postgres listen example with a small chat example [@JockeM]
- [#2602]: Support naming migrations sequentially [@vmax]
- [#2634]: Adding PgHasArrayType for &[u8;N] [@snf]
- [#2646]: Support for setting client certificate and key from bytes [@wyhaya]
- [#2664]: Automatically infer migration type [@vmax]
- [#2712]: Add impl for
Type
,Decode
, andEncode
forBox<str>
andBox<[u8]>
[@grant0417]
- [#2650]: Cleanup format arguments [@nyurik]
- [#2695]: remove &mut PoolConnection from Executor docs [@olback]
- This impl was removed in 0.7.0 because of coherence issues.
- [#2706]: Clarify where optional features should be enabled [@kryptan]
- [#2717]: Update README.md [@fermanjj]
- [#2739]: Bump mariadb CI images + mysql unpin [@grooverdan]
- [#2742]: Implemented poll_flush for Box<S:Socket> [@bobozaur]
- [#2740]: Remove sealed trait comments from documentation [@bobozaur]
- [#2750]: Fix #2384, bump flume to v0.11.0 [@madadam]
- [#2757]: Remove unused
remove_dir_all
crate fromsqlx-cli
, fixes RUSTSEC-2023-0018 [@aldur]
- [#2624]: Documentation typo: BYTE -> BINARY [@sebastianv89]
- [#2628]: docs: 0.7 is stable in the entire README [@marcusirgens]
- [#2630]: fix(postgres): fix buffer management in PgCopyIn::read_from [@tsing]
- [#2651]: Chore: Fix few build warnings, and make CI fail on warn [@nyurik]
- [#2670]: fix: ignore extra fields in Postgres describe parsing [@abonander]
- [#2687]: docs: Fix description of
min_connections
[@hakoerber]
This release mainly addresses issues reported with the 0.7.0 release.
16 pull requests were merged this release cycle.
- [#2551]: Introduce build_query_scalar for QueryBuilder [@iamquang95]
- [#2605]: Implement Default for QueryBuilder [@Xydez]
- [#2616]: feat(sqlx-core): add table function to database error [@saiintbrisson]
- [#2619]: feat: allow opt-out of
PgHasArrayType
with#[derive(sqlx::Type)]
[@abonander]- TL;DR: if you're getting errors from
#[derive(sqlx::Type)]
with#[sqlx(transparent)]
regardingPgHasArrayType
not being implemented, add#[sqlx(no_pg_array)]
to fix.
- TL;DR: if you're getting errors from
- [#2566]: improve docs about migration files [@jnnnnn]
- [#2576]: Major Version Update clap to 4.0 [@titaniumtraveler]
- [#2597]: Bump webpki-roots to v0.24 [@paolobarbolini]
- [#2603]: docs(changelog): be more verbose about offline mode breaking change [@mrl5]
- [#2553]: Implement
Clone
forPoolOptions
manually (#2548) [@alilleybrinker] - [#2580]: Update README.md now that 0.7.0 is no longer in alpha [@saolof]
- [#2585]: Fix for Issue #2549 - cannot use feature "rust_decimal" without also using "bigdecimal" [@deneut]
- [#2586]: Fix optional dependency on sqlx-macros [@kitterion]
- [#2593]: Correct mention of the
tls-native-tls
in the documentation. [@denschub] - [#2599]: Remove incorrect CAST in test database cleanup for MySQL. [@fd]
- [#2613]: Fix readme.md to reduce confusion about optional features (decimal->rust_decimal) [@vabka]
- [#2620]: fix(sqlite/any): encode bool as integer [@saiintbrisson]
At least 70 pull requests were merged this release cycle! (The exact count is muddied with pull requests for alpha releases and such.) And we gained 43 new contributors! Thank you to everyone who helped make this release a reality.
Many revisions were made to query analysis in the SQLite driver; these are all potentially breaking changes
as they can change the output of sqlx::query!()
et al. We'd like to thank [@tyrelr] for their numerous PRs to
this area.
The MSSQL driver has been removed as it was not nearly at the same maturity level as the other drivers. As previously announced, we have plans to introduce a fully featured replacement as a premium offering, alongside drivers for other proprietary databases, with the goal to support full-time development on SQLx.
If interested, please email your inquiry to [email protected].
The offline mode for the queries has been changed to use a separate file per query!()
invocation,
which is intended to reduce the number of conflicts when merging branches in a project that both modified queries.
This means that CLI flag --merged
is no longer supported. See [#2363] for details and make sure that your
sqlx-cli
version is in sync with the sqlx
version in your project.
The type ascription override syntax for the query macros has been deprecated,
as parse support for it has been removed in syn 2.0
, which we'll be upgrading to in the next breaking release.
This can be replaced with type overrides using casting syntax (as
).
See [#2483] for details.
- [#1946]: Fix compile time verification performance regression for sqlite [@liningpan]
- [#1960]: Fix sqlite update return and order by type inference [@tyrelr]
- [#1984]: Sqlite EXPLAIN type inference improvements [@rongcuid]
- [#2039]: Break drivers out into separate crates, clean up some technical debt [@abonander]
- All deprecated items have been removed.
- The
mssql
feature and associated database driver has been deleted from the source tree. It will return as part of our planned SQLx Pro offering as a from-scratch rewrite with extra features (such as TLS) and type integrations that were previously missing. - The
runtime-actix-*
features have been deleted. They were previously changed to be aliases of theirruntime-tokio-*
counterparts for backwards compatibility reasons, but their continued existence is misleading as SQLx has no special knowledge of Actix anymore.- To fix, simply replace the
runtime-actix-*
feature with itsruntime-tokio-*
equivalent.
- To fix, simply replace the
- The
git2
feature has been removed. This was a requested integration from a while ago that over time made less and less sense to be part of SQLx itself. We have to be careful with the crates we add to our public API as each one introduces yet another semver hazard. The expected replacement is to make#[derive(sqlx::Type)]
useful enough that users can write wrapper types for whatever they want to use without SQLx needing to be specifically aware of it. - The
Executor
impls forTransaction
andPoolConnection
have been deleted because they cannot exist in the new crate architecture without rewriting theExecutor
trait entirely.- To fix this breakage, simply add a dereference where an
impl Executor
is expected, as they both dereference to the inner connection type which will still implement it:&mut transaction
->&mut *transaction
&mut connection
->&mut *connection
- These cannot be blanket impls as it triggers an overflow in the compiler due to the lack of lazy normalization, and the driver crates cannot provide their own impls due to the orphan rule.
- We're expecting to do another major refactor of traits to incorporate generic associated types (GAT). This will mean another major release of SQLx but ideally most API usage will not need to change significantly, if at all.
- To fix this breakage, simply add a dereference where an
- The fields of
Migrator
are now#[doc(hidden)]
and semver-exempt; they weren't meant to be public. - The
offline
feature has been removed from thesqlx
facade crate and is enabled unconditionally as most users are expected to have enabled it anyway and disabling it doesn't seem to appreciably affect compile times. - The
decimal
feature has been renamed torust_decimal
to match the crate it actually provides integrations for. AnyDriver
andAnyConnection
now require eithersqlx::any::install_drivers()
orsqlx::any::install_default_drivers()
to be called at some point during the process' lifetime before the first connection is made, as the set of possible drivers is now determined at runtime. This was determined to be the least painful way to provide knowledge of database drivers toAny
without them being hardcoded.- The
AnyEncode
trait has been removed.
- [#2109]: feat: better database errors [@saiintbrisson]
- [#2094]: Update libsqlite3-sys to 0.25.1 [@penberg]
- Alongside this upgrade, we are now considering the linkage to
libsqlite3-sys
to be semver-exempt, and we reserve the right to upgrade it as necessary. If you are usinglibsqlite3-sys
directly or a crate that links it such asrusqlite
, you should pin the versions of both crates to avoid breakages fromcargo update
:
- Alongside this upgrade, we are now considering the linkage to
[dependencies]
sqlx = { version = "=0.7.0", features = ["sqlite"] }
rusqlite = "=0.29.0"
- [#2132]: fix: use owned Builder pattern for ConnectOptions [@ar3s3ru]
- [#2253]: Sqlite describe fixes [@tyrelr]
- [#2285]:
time
: Assume UTC when decoding a DATETIME column in sqlite [@nstinus] - [#2363]: [offline] Change prepare to one-file-per-query [@cycraig]
- [#2387]: PATCH: bump libsqlite3-sys to patched version [@grantkee]
- [#2409]: fix(#2407): respect the HaltIfNull opcode when determining nullability [@arlyon]
- [#2459]: limit the number of instructions that can be evaluated [@tyrelr]
- [#2467]: Add and improve sqlite describe performance benchmarks [@tyrelr]
- [#2491]: sqlite date macro support [@Arcayr]
- Changes
OffsetDateTime
to be the first type used when deserializing atimestamp
type.
- Changes
- [#2496]: Bump to libsqlite3-sys 0.26 [@mdecimus]
- [#2508]: Sqlite analytical [@tyrelr]
- [#1850]: Add client SSL authentication using key-file for Postgres, MySQL and MariaDB [@ThibsG]
- [#2088]: feat: Add set_connect_options method to Pool [@moatra]
- [#2113]: Expose PoolOptions for reading [@FSMaxB]
- [#2115]: Allow using complex types in
try_from
when derivingFromRow
[@95ulisse] - [#2116]: [SQLite] Add option to execute
PRAGMA optimize;
on close of a connection [@miles170] - [#2189]: Added regexp support in sqlite [@VictorKoenders]
- [#2224]: Add From impls for Json [@dbeckwith]
- [#2256]: add progress handler support to sqlite [@nbaztec]
- [#2366]: Allow ignoring attributes for deriving FromRow [@grgi]
- [#2369]: new type support in query_as [@0xdeafbeef]
- [#2379]: feat: add
Connection::shrink_buffers
,PoolConnection::close
[@abonander] - [#2400]: fix(docs): example of
sqlx_macros_unstable
in config.toml [@df51d] - [#2469]: Add Simple format for Uuid for MySQL & SQLite. [@MidasLamb]
- [#2483]: chore: add deprecation notice for type ascription use [@saiintbrisson]
- [#2506]: add args to query builder (#2494) [@cemoktra]
- [#2554]: Impl
AsMut
for advisory lock types (#2520) [@alilleybrinker] - [#2559]: Add CLI autocompletion using clap_complete [@titaniumtraveler]
- [#2185]: Initial work to switch to
tracing
[@CosmicHorrorDev] - [#2193]: Start testing on Postgres 15 and drop Postgres 10 [@paolobarbolini]
- We reserve the right to drop support for end-of-lifed database versions as discussed in our FAQ.
- [#2213]: Use
let else
statements in favor of macro [@OverHash] - [#2365]: Update dependencies [@paolobarbolini]
- [#2371]: Disable rustls crate logging feature by default up to date [@sergeiivankov]
- [#2373]: chore: Use tracing's fields to get structured logs [@jaysonsantos]
- [#2393]: Lower default logging level for statements to Debug [@bnoctis]
- [#2445]: Traverse symlinks when resolving migrations [@tgeoghegan]
- [#2485]: chore(sqlx-postgres): replace
dirs
withhome
&etcetera
[@utkarshgupta137] - [#2515]: Bump mac_address to 1.1.5 [@repnop]
- [#2440]: Update rustls to 0.21, webpki-roots to 0.23 [@SergioBenitez]
- [#2563]: Update rsa to 0.9 [@paolobarbolini]
- [#2564]: Update bitflags to v2 [@paolobarbolini]
- [#2565]: Bump indexmap and ahash [@paolobarbolini]
- [#2574]: doc: make it clear that
ConnectOptions
types implFromStr
[@abonander]
- [#2098]: Fix sqlite compilation [@cycraig]
- [#2120]: fix logical merge conflict [@tyrelr]
- [#2133]: Postgres OID resolution query does not take into account current
search_path
[@95ulisse] - [#2156]: Fixed typo. [@cdbfoster]
- [#2179]: fix: ensures recover from fail with PgCopyIn [@andyquinterom]
- [#2200]: Run CI on *-dev branch [@joehillen]
- [#2222]: Add context to confusing sqlx prepare parse error [@laundmo]
- [#2271]: feat: support calling Postgres procedures with the macros [@bgeron]
- [#2282]: Don't run EXPLAIN nullability analysis on Materialize [@benesch]
- [#2319]: Set whoami default-features to false [@thedodd]
- [#2352]: Preparing 0.7.0-alpha.1 release [@abonander]
- [#2355]: Fixed the example code for
sqlx::test
[@kenkoooo] - [#2367]: Fix sqlx-cli create, drop, migrate [@cycraig]
- [#2376]: fix(pool): close when last handle is dropped, extra check in
try_acquire
[@abonander] - [#2378]: Fix README build badge [@dbrgn]
- [#2398]: fix(prepare): store temporary query files inside the workspace [@aschey]
- [#2402]: fix: drop old time 0.1.44 dep [@codahale]
- [#2413]: fix(macros-core): use of undeclared
tracked_path
[@df51d] - [#2420]: Enable runtime-tokio feature of sqlx when building sqlx-cli [@paolobarbolini]
- [#2453]: in README.md, correct spelling and grammar [@vizvasrj]
- [#2454]: fix: ensure fresh test db's aren't accidentally deleted by do_cleanup [@phlip9]
- [#2507]: Exposing the Oid of PostgreSQL types [@Razican]
- [#2519]: Use ::std::result::Result::Ok in output.rs [@southball]
- [#2569]: Fix broken links to mysql error documentation [@titaniumtraveler]
- [#2570]: Add a newline to the generated JSON files [@nyurik]
- [#2572]: Do not panic when
PrepareOk
fails to decode [@stepantubanov] - [#2573]: fix(sqlite) Do not drop notify mutex guard until after condvar is triggered [@andrewwhitehead]
This is a hotfix to address the breakage caused by transitive dependencies upgrading to syn = "2"
.
We set default-features = false
for our dependency on syn = "1"
to be good crates.io citizens,
but failed to enable the features we actually used, which went undetected because we transitively depended on
syn
with the default features enabled through other crates,
and so they were also on for us because features are additive.
When those other dependencies upgraded to syn = "2"
it was no longer enabling those features for us,
and so compilation broke for projects that don't also depend on syn = "1"
, transitively or otherwise.
There is no PR for this fix as there was no longer a dedicated development branch for 0.6
,
but discussion can be found in issue #2418.
As of this release, the 0.7
release is in alpha and so development is no longer occurring against 0.6
.
This fix will be forward-ported to 0.7
.
25 pull requests were merged this release cycle.
- [#1081]: Add
try_from
attribute forFromRow
derive [@zzhengzhuo]- Exemplifies "out of sight, out of mind." It's surprisingly easy to forget about PRs when they get pushed onto the second page. We'll be sure to clean out the backlog for 0.7.0.
- [#2014]: Support additional SQLCipher options in SQLite driver. [@szymek156]
- [#2052]: Add issue templates [@abonander]
- [#2053]: Add documentation for
IpAddr
support in Postgres [@rakshith-ravi] - [#2062]: Add extension support for SQLite [@bradfier]
- [#2063]: customizable db locking during migration [@fuzzbuck]
- [#2025]: Bump sqlformat to 2.0 [@NSMustache]
- [#2056]: chore: Switch to sha1 crate [@stoically]
- [#2071]: Use cargo check consistently in
prepare
[@cycraig]
- [#1991]: Ensure migration progress is not lost for Postgres, MySQL and SQLite. [@crepererum]
- [#2023]: Fix expansion of
#[sqlx(flatten)]
forFromRow
derive [@RustyYato] - [#2028]: Use fully qualified path when forwarding to
#[test]
from#[sqlx::test]
[@alexander-jackson] - [#2040]: Fix typo in
FromRow
docs [@zlidner] - [#2046]: added flag for PIPES_AS_CONCAT connection setting for MySQL to fix #2034 [@marcustut]
- [#2055]: Use unlock notify also on
sqlite3_exec
[@madadam] - [#2057]: Make begin,commit,rollback cancel-safe in sqlite [@madadam]
- [#2058]: fix typo in documentation [@lovasoa]
- [#2067]: fix(docs): close code block in query_builder.rs [@abonander]
- [#2069]: Fix
prepare
race condition in workspaces [@cycraig]\- NOTE: this changes the directory structure under
target/
thatcargo sqlx prepare
depends on. If you use offline mode in your workflow, please reruncargo install sqlx-cli
to upgrade.
- NOTE: this changes the directory structure under
- [#2072]: SqliteConnectOptions typo [@fasterthanlime]
- [#2074]: fix: mssql uses unsigned for tinyint instead of signed [@he4d]
- [#2081]: close unnamed portal after each executed extended query [@DXist]
- [#2086]: PgHasArrayType for transparent types fix. [@Wopple]
- NOTE: this is a breaking change and has been postponed to 0.7.0.
- [#2089]: fix: Remove default chrono dep on time for sqlx-cli [@TravisWhitehead]
- [#2091]: Sqlite explain plan log efficiency [@tyrelr]
33 pull requests were merged this release cycle.
- [#1495]: Add example for manual implementation of the
FromRow
trait [@Erik1000] - [#1822]: (Postgres) Add support for
std::net::IpAddr
[@meh]- Decoding returns an error if the
INET
value in Postgres is a prefix and not a full address (/32
for IPv4,/128
for IPv6).
- Decoding returns an error if the
- [#1865]: Add SQLite support for the
time
crate [@johnbcodes] - [#1902]: Add an example of how to use
QueryBuilder::separated()
[@sbeckeriv] - [#1917]: Added docs for
sqlx::types::Json
[@jayy-lmao] - [#1919]: Implement
Clone
forPoolOptions
[@Thomasdezeeuw] - [#1953]: Support Rust arrays in Postgres [@e00E]
- [#1954]: Add
push_tuples
forQueryBuilder
[@0xdeafbeef] - [#1959]: Support
#[sqlx(flatten)]
attribute inFromRow
[@TheoOiry] - [#1967]: Add example with external query files [@JoeyMckenzie]
- [#1985]: Add
query_builder::Separated::push_bind_unseparated()
[@0xdeafbeef] - [#2001]: Implement
#[sqlx::test]
for general use- Includes automatic database management, migration and fixture application.
- Drops support for end-of-lifed database versions, see PR for details.
- [#2005]:
QueryBuilder
improvements [@abonander]- Raw SQL getters, new method to build
QueryAs
instead ofQuery
.
- Raw SQL getters, new method to build
- [#2013]: (SQLite) Allow VFS to be set as URL query parameter [@liningpan]
- [#1679]: refactor: alias actix-* features to their equivalent tokio-* features [@robjtede]
- [#1906]: replaced all uses of "uri" to "url" [@RomainStorai]
- [#1965]: SQLite improvements [@abonander]
- [#1977]: Docs: clarify relationship between
query_as!()
andFromRow
[@abonander] - [#2003]: Replace
dotenv
withdotenvy
[@abonander]
- [#1802]: Try avoiding a full clean in
cargo sqlx prepare --merged
[@LovecraftianHorror] - [#1848]: Fix type info access in
Any
database driver [@raviqqe] - [#1910]: Set
CARGO_TARGET_DIR
when compiling queries [@sedrik] - [#1915]: Pool: fix panic when using callbacks [@abonander]
- [#1930]: Don't cache SQLite connection for macros [@LovecraftianHorror]
- [#1948]: Fix panic in Postgres
BYTEA
decode [@e00E] - [#1955]: Fix typo in FAQ [@kenkoooo]
- [#1968]: (Postgres) don't panic if
S
orV
notice fields are not UTF-8 [@abonander] - [#1969]: Fix sqlx-cli build [@ivan]
- [#1974]: Use the
rust-cache
action for CI [@abonander] - [#1988]: Agree on a single default runtime for the whole workspace [@crepererum]
- [#1989]: Fix panics in
PgListener
[@crepererum] - [#1990]: Switch
master
tomain
in docs [@crepererum]- The change had already been made in the repo, the docs were out of date.
- [#1993]: Update versions in quickstart examples in README [@UramnOIL]
This release marks the end of the 0.5.x series of releases and contains a number of breaking changes, mainly to do with backwards-incompatible dependency upgrades.
As we foresee many more of these in the future, we surveyed the community on how to handle this; the consensus appears to be "just release breaking changes more often."
As such, we expect the 0.6.x release series to be a shorter one.
39 pull requests(!) (not counting "prepare 0.5.12 release", of course) were merged this release cycle.
- [#1384]: (Postgres) Move
server_version_num
from trait to inherent impl [@AtkinsChang] - [#1426]: Bump
ipnetwork
to 0.19 [@paolobarbolini] - [#1455]: Upgrade
time
to 0.3 [@paolobarbolini] - [#1505]: Upgrade
rustls
to 0.20 [@paolobarbolini]- Fortunately, future upgrades should not be breaking as
webpki
is no longer exposed in the API.
- Fortunately, future upgrades should not be breaking as
- [#1529]: Upgrade
bigdecimal
to 0.3 [@e00E] - [#1602]: postgres: use
Oid
everywhere instead ofu32
[@paolobarbolini]- This drops the
Type
,Decode
,Encode
impls foru32
for Postgres as it was misleading. Postgres doesn't support unsigned ints without using an extension. These impls were decoding PostgresOID
s as bareu32
s without any context (and trying to bind au32
to a query would produce anOID
value in SQL). This changes that to use a newtype instead, for clarity.
- This drops the
- [#1612]: Make all
ConnectOptions
types cloneable [@05storm26] - [#1618]: SQLite
chrono::DateTime<FixedOffset>
timezone fix [@05storm26]DateTime<FixedOffset>
will be stored in SQLite with the correct timezone instead of always in UTC. This was flagged as a "potentially breaking change" since it changes how dates are sent to SQLite.
- [#1733]: Update
git2
to 0.14 [@joshtriplett] - [#1734]: Make
PgLTree::push()
infallible and takePgLTreeLabel
directly [@sebpuetz] - [#1785]: Fix Rust type for SQLite
REAL
[@pruthvikar]- Makes the macros always map a
REAL
column tof64
instead off32
as SQLite uses only 64-bit floats.
- Makes the macros always map a
- [#1816]: Improve SQLite support for sub-queries and CTEs [@tyrelr]
- This likely will change the generated code for some invocations
sqlx::query!()
with SQLite.
- This likely will change the generated code for some invocations
- [#1821]: Update
uuid
crate to v1 [@paolobarbolini] - [#1901]: Pool fixes and breaking changes [@abonander]
- Renamed
PoolOptions::connect_timeout
toacquire_timeout
for clarity. - Changed the expected signatures for
PoolOptions::after_connect
,before_acquire
,after_release
- Changed the signature for
Pool::close()
slightly- Now eagerly starts the pool closing,
.await
ing is only necessary if you want to ensure a graceful shutdown.
- Now eagerly starts the pool closing,
- Deleted
PoolConnection::release()
which was previously deprecated in favor ofPoolConnection::detach()
. - Fixed connections getting leaked even when calling
.close()
.
- Renamed
- [#1748]: Derive
PgHasArrayType
for#[sqlx(transparent)]
types [@carols10cents]- This change was released with 0.5.12 but we didn't realize it was a breaking change at the time.
It was reverted in 0.5.13 and postponed until this release.
- This change was released with 0.5.12 but we didn't realize it was a breaking change at the time.
- [#1843]: Expose some useful methods on
PgValueRef
[@mfreeborn] - [#1889]: SQLx-CLI: add
--connect-timeout
[@abonander]- Adds a default 10 second connection timeout to all commands.
- [#1890]: Added test for mssql LoginAck [@walf443]
- [#1891]: Added test for mssql ProtocolInfo [@walf443]
- [#1892]: Added test for mssql ReturnValue [@walf443]
- [#1895]: Add support for
i16
toAny
driver [@EthanYuan] - [#1897]: Expose
ConnectOptions
andPoolOptions
onPool
and database name onPgConnectOptions
[@Nukesor]
- [#1782]: Reuse a cached DB connection instead of always opening a new one for
sqlx-macros
[@LovecraftianHorror] - [#1807]: Bump remaining dependencies [@paolobarbolini]
- [#1808]: Update to edition 2021 [@paolobarbolini]
- Note that while SQLx does not officially track an MSRV and only officially supports the latest stable Rust, this effectively places a lower bound of 1.56.0 on the range of versions it may work with.
- [#1823]: (sqlx-macros) Ignore deps when getting metadata for workspace root [@LovecraftianHorror]
- [#1831]: Update
crc
to 3.0 [@djc] - [#1887]: query_as: don't stop stream after decoding error [@lovasoa]
- [#1814]: SQLx-cli README: move
Usage
to the same level asInstall
[@tobymurray] - [#1815]: SQLx-cli README: reword "building in offline mode" [@tobymurray]
- [#1818]: Trim
[]
from host string before passing to TcpStream [@smonv]- This fixes handling of database URLs with IPv6 hosts.
- [#1842]: Fix usage of
serde_json
in macros [@mfreeborn] - [#1855]: Postgres: fix panics on unknown type OID when decoding [@demurgos]
- [#1856]: MySQL: support COLLATE_UTF8MB4_0900_AI_CI [@scottwey]
- Fixes the MySQL driver thinking text columns are bytestring columns when querying against a Planetscale DB.
- [#1861]: MySQL: avoid panic when streaming packets are empty [@e-rhodes]
- [#1863]: Fix nullability check for inner joins in Postgres [@OskarPersson]
- [#1881]: Fix
field is never read
warnings on Postgres test [@walf443] - [#1882]: Fix
unused result must be used
warnings [@walf443] - [#1888]: Fix migration checksum comparison during
sqlx migrate info
[@mdtusz] - [#1894]: Fix typos [@kianmeng]
This is a hotfix that reverts #1748 as that was an accidental breaking change:
the generated PgHasArrayType
impl conflicts with manual impls of the trait.
This change will have to wait for 0.6.0.
27 pull requests were merged this release cycle.
- [#1641]: Postgres: Convenient wrapper for advisory locks [@abonander]
- [#1675]: Add function to undo migrations [@jdrouet]
- [#1722]: Postgres: implement
PgHasArrayType
forserde_json::{Value, RawValue}
[@abreis] - [#1736]: Derive
Clone
forMySqlArguments
andMssqlArguments
[@0xdeafbeef] - [#1748]: Derive
PgHasArrayType
for#[sqlx(transparent)]
types [@carols10cents] - [#1754]: Include affected rows alongside returned rows in query logging [@david-mcgillicuddy-moixa]
- [#1757]: Implement
Type
forCow<str>
for MySQL, MSSQL and SQLite [@ipetkov] - [#1769]: sqlx-cli: add
--source
to migration subcommands [@pedromfedricci] - [#1774]: Postgres: make
extra_float_digits
settable [@abonander]- Can be set to
None
for Postgres or third-party database servers that don't support the option.
- Can be set to
- [#1776]: Implement close-event notification for Pool [@abonander]
- Also fixes
PgListener
preventingPool::close()
from resolving.
- Also fixes
- [#1780]: Implement query builder [@crajcan]
- See also [#1790]: Document and expand query builder [@abonander]
- [#1781]: Postgres: support
NUMERIC[]
usingdecimal
feature [@tm-drtina] - [#1784]: SQLite: add
FromStr
,Copy
,PartialEq
,Eq
impls for options enums [@andrewwhitehead]
- [#1625]: Update RustCrypto crates [@paolobarbolini]
- [#1725]: Update
heck
to 0.4 [@paolobarbolini] - [#1738]: Update
regex
[@Dylan-DPC] - [#1763]: SQLite: update
libsqlite3-sys
[@espindola]
- [#1719]: Fix a link in
query!()
docs [@vbmade2000] - [#1731]: Postgres: fix option passing logic [@liushuyu]
- [#1735]: sqlx-cli: pass
DATABASE_URL
to command spawned inprepare
[@LovecraftianHorror] - [#1741]: Postgres: fix typo in
TSTZRANGE
[@mgrachev] - [#1761]: Fix link from
QueryAs
toquery_as()
in docs [@mgrachev] - [#1786]: MySQL: silence compile warnings for unused fields [@andrewwhitehead]
- [#1789]: SQLite: fix left-joins breaking
query!()
macros [@tyrelr] - [#1791]: Postgres: fix newline parsing of
.pgpass
files [@SebastienGllmt] - [#1799]:
PoolConnection
: don't leak connection permit if drop task fails to run [@abonander]
20 pull requests were merged this release cycle.
- [#1610]: Allow converting
AnyConnectOptions
to a specificConnectOptions
[@05storm26] - [#1652]: Implement
From
forAnyConnection
[@genusistimelord] - [#1658]: Handle
SQLITE_LOCKED
[@madadam] - [#1665]: Document offline mode usage with feature flags [@sedrik]
- [#1680]: Show checksum mismatches in
sqlx migrate info
[@ifn3] - [#1685]: Add tip for setting
opt-level
forsqlx-macros
[@LovecraftianHorror] - [#1687]: Docs:
Acquire
examples and alternative [@stoically] - [#1696]: Postgres: support for
ltree
[@cemoktra] - [#1710]: Postgres: support for
lquery
[@cemoktra]
- [#1605]: Remove unused dependencies [@paolobarbolini]
- [#1606]: Add target context to Postgres
NOTICE
logs [@dbeckwith] - [#1684]: Macros: Cache parsed
sqlx-data.json
instead of reparsing [@LovecraftianHorror]
- [#1608]: Drop worker shared state in shutdown (SQLite) [@andrewwhitehead]
- [#1619]: Docs(macros): remove sentences banning usage of
as _
[@k-jun] - [#1626]: Simplify
cargo-sqlx
command-line definition [@tranzystorek-io] - [#1636]: Fix and extend Postgres transaction example [@taladar]
- [#1657]: Fix typo in macro docs [@p9s]
- [#1661]: Fix binding
Option<T>
forAny
driver [@ArGGu] - [#1667]: MySQL: Avoid panicking if packet is empty [@nappa85]
- [#1692]: Postgres: Fix power calculation when encoding
BigDecimal
intoNUMERIC
[@VersBinarii]
Additionally, we have introduced two mitigations for the issue of the cyclic dependency on ahash
:
- We re-downgraded our version requirement on
indexmap
from1.7.0
back to1.6.2
so users can pin it to that version as recommended in aHash#95. - Thanks to the work of @LovecraftianHorror in #1684, we no longer require the
preserve_order
feature ofserde_json
which gives users another place to break the cycle by simply not enabling that feature.- This may introduce extra churn in Git diffs for
sqlx-data.json
, however. If this is an issue for you but the dependency cycle isn't, you can re-enable thepreserve_order
feature:
[dependencies] serde_json = { version = "1", features = ["preserve_order"] }
- This may introduce extra churn in Git diffs for
A whopping 31 pull requests were merged this release cycle!
According to this changelog, we saw 18 new contributors! However, some of these folks may have missed getting mentioned in previous entries since we only listed highlights. To avoid anyone feeling left out, I put in the effort this time and tried to list every single one here.
- [#1228]: Add
Pool::any_kind()
[@nitnelave] - [#1343]: Add
Encode/Decode
impl forCow<'_, str>
[@Drevoed] - [#1474]: Derive
Clone
,Copy
forAnyKind
[@yuyawk] - [#1497]: Update FAQ to explain how to configure docs.rs to build a project using SQLx [@russweas]
- [#1498]: Add description of migration file structure to
migrate!()
docs [@zbigniewzolnierowicz] - [#1508]: Add
.persistent(bool)
toQueryAs
,QueryScalar
[@akiradeveloper] - [#1514]: Add support for serialized threading mode to SQLite [@LLBlumire]
- [#1523]: Allow
rust_decimal::Decimal
inPgRange
[@meh] - [#1539]: Support
PGOPTIONS
and adding custom configuration options inPgConnectOptions
[@liushuyu] - [#1562]: Re-export
either::Either
used byExecutor::fetch_many()
[@DoumanAsh] - [#1584]: Add feature to use RusTLS instead of
native-tls
forsqlx-cli
[@SonicZentropy] - [#1592]: Add
AnyConnection::kind()
[@05storm26]
- [#1385]: Rewrite Postgres array handling to reduce boilerplate and allow custom types [@jplatte]
- [#1479]: Remove outdated mention of
runtime-async-std-native-tls
as the default runtime in README.md [@yerke] - [#1526]: Revise
Pool
docs in a couple places [@abonander] - [#1535]: Bump
libsqlite-sys
to0.23.1
[@nitsky] - [#1551]: SQLite: make worker thread responsible for all FFI calls [@abonander]
- If you were encountering segfaults with the SQLite driver, there's a good chance this will fix it!
- [#1557]: CI: test with Postgres 14 [@paolobarbolini]
- [#1571]: Make
whoami
dep optional, only pull it in for Postgres [@joshtriplett] - [#1572]: Update
rsa
crate to 0.5 [@paolobarbolini] - [#1591]: List SeaORM as an ORM option in the README [@kunjee17]
- [#1601]: Update
itoa
anddirs
[@paolobarbolini]
- [#1475]: Fix panic when converting a negative
chrono::Duration
toPgInterval
[@yuyawk] - [#1483]: Fix error when decoding array of custom types from Postgres [@demurgos
- [#1501]: Reduce
indexmap
version requirement to1.6.2
[@dimfeld] - [#1511]: Fix element type given to Postgres for arrays of custom enums [@chesedo]
- [#1517]: Fix mismatched type errors in MySQL type tests [@abonander]
- [#1537]: Fix missing re-export of
PgCopyIn
[@akiradeveloper] - [#1566]: Match
~/.pgpass
password after URL parsing and fix user and database ordering [@D1plo1d] - [#1582]:
cargo sqlx prepare
: Append to existingRUSTFLAGS
instead of overwriting [@tkintscher] - [#1587]: SQLite: if set, send
PRAGMA key
on a new connection before anything else. [@parazyd]- This should fix problems with being unable to open databases using SQLCipher.
A hotfix release to address the issue of the sqlx
crate itself still depending on older versions of sqlx-core
and
sqlx-macros
.
No other changes from 0.5.8
.
A total of 24 pull requests were merged this release cycle! Some highlights:
- [#1289] Support the
immutable
option on SQLite connections [@djmarcin] - [#1295] Support custom initial options for SQLite [@ghassmo]
- Allows specifying custom
PRAGMA
s and overriding those set by SQLx.
- Allows specifying custom
- [#1345] Initial support for Postgres
COPY FROM/TO
[@montanalow, @abonander] - [#1439] Handle multiple waiting results correctly in MySQL [@eagletmt]
- [#1392] use
resolve_path
when getting path forinclude_str!()
[@abonander]- Fixes a regression introduced by [#1332].
- [#1393] avoid recursively spawning tasks in
PgListener::drop()
[@abonander]- Fixes a panic that occurs when
PgListener
is dropped inasync fn main()
.
- Fixes a panic that occurs when
A large bugfix release, including but not limited to:
- [#1329] Implement
MACADDR
type for Postgres [@nomick] - [#1363] Fix
PortalSuspended
for array of composite types in Postgres [@AtkinsChang] - [#1320] Reimplement
sqlx::Pool
internals usingfutures-intrusive
[@abonander]- This addresses a number of deadlocks/stalls on acquiring connections from the pool.
- [#1332] Macros: tell the compiler about external files/env vars to watch [@abonander]
- Includes
sqlx build-script
to create abuild.rs
to watchmigrations/
for changes. - Nightly users can try
RUSTFLAGS=--cfg sqlx_macros_unstable
to tell the compiler to watchmigrations/
for changes instead of using a build script. - See the new section in the docs for
sqlx::migrate!()
for details.
- Includes
- [#1351] Fix a few sources of segfaults/errors in SQLite driver [@abonander]
- [#1323] Keep track of column typing in SQLite EXPLAIN parsing [@marshoepial]
- This fixes errors in the macros when using
INSERT/UPDATE/DELETE ... RETURNING ...
in SQLite.
- This fixes errors in the macros when using
A total of 25 pull requests were merged this release cycle!
-
[#1211] Even more tweaks and fixes to the Pool internals [@abonander]
-
[#1213] Add support for bytes and
chrono::NaiveDateTime
toAny
[@guylapid] -
[#1224] Add support for
chrono::DateTime<Local>
toAny
withMySQL
[@NatPRoach] -
[#1216] Skip empty lines and comments in pgpass files [@feikesteenbergen]
-
[#1218] Add support for
PgMoney
to the compile-time type-checking [@iamsiddhant05]
-
[#1149] Tweak and optimize Pool internals [@abonander]
-
[#1132] Remove
'static
bound onConnection::transaction
[@argv-minus-one] -
[#1099] [#1097] Truncate buffer when
BufStream
is dropped [@Diggsey]
-
[#1170] Remove
Self: Type
bounds inEncode
/Decode
implementations for arrays [@jplatte]Enables working around the lack of support for user-defined array types:
#[derive(sqlx::Encode)] struct Foos<'a>(&'a [Foo]); impl sqlx::Type<sqlx::Postgres> for Foos<'_> { fn type_info() -> PgTypeInfo { PgTypeInfo::with_name("_foo") } } query_as!( Whatever, "<QUERY with $1 of type foo[]>", Foos(&foo_vec) as _, )
-
[#1141] Use
u16::MAX
instead ofi16::MAX
for a check against the largest number of parameters in a query [@crajcan] -
[#1100] Explicitly
UNLISTEN
before returning connections to the pool inPgListener
[@Diggsey]
-
[#1161] Catch
SQLITE_MISUSE
on connection close and panic [@link2xt] -
[#1160] Do not cast pointers to
i32
(cast tousize
) [@link2xt] -
[#1156] Reset the statement when
fetch_many
stream is dropped [@link2xt]
- Update sqlx-rt to 0.3.
-
[#983] [#1022] Upgrade async runtime dependencies [@seryl, @ant32, @jplatte, @robjtede]
- tokio 1.0
- actix-rt 2.0
-
[#854] Allow chaining
map
andtry_map
[@jplatte]Additionally enables calling these combinators with the macros:
let ones: Vec<i32> = query!("SELECT 1 as foo") .map(|row| row.foo) .fetch_all(&mut conn).await?;
-
[#940] Rename the
#[sqlx(rename)]
attribute used to specify the type name on the database side to#[sqlx(type_name)]
[@jplatte]. -
[#976] Remove the
Done
trait. The.rows_affected()
method is now available as an inherent method onPgQueryResult
,MySqlQueryResult
and so on. [@jplatte] -
[#1007] Remove
any::AnyType
(and replace with directly implementingType<Any>
) [@jplatte]
-
[#919] For SQLite, add support for unsigned integers [@dignifiedquire]
-
[#1002] For SQLite,
GROUP BY
inquery!
caused an infinite loop at compile time. [@pymongo] -
[#979] For MySQL, fix support for non-default authentication. [@sile]
-
[#918] Recover from dropping
wait_for_conn
inside Pool. [@antialize]
-
[#908] Fix
whoami
crash on FreeBSD platform [@fundon] [@AldaronLau] -
[#895] Decrement pool size when connection is released [@andrewwhitehead]
-
[#878] Fix
conn.transaction
wrapper [@hamza1311]conn.transaction(|transaction: &mut Transaction<Database> | { // ... });
-
[#860] Add
rename_all
toFromRow
and addcamelCase
andPascalCase
[@framp] -
[#839] Add (optional) support for
bstr::BStr
,bstr::BString
, andgit2::Oid
[@joshtriplett]
-
[#852] Fix potential 100% CPU usage in
fetch_one
/fetch_optional
[@markazmierczak] -
[#850] Add
synchronous
option toSqliteConnectOptions
[@markazmierczak]
-
[#876] Add support for
BYTEA[]
to compile-time type-checking [@augustocdias]
Fix docs.rs build by enabling a runtime feature in the docs.rs metadata in Cargo.toml
.
-
[#774] Fix usage of SQLx derives with other derive crates [@NyxCode]
-
[#762] Fix
migrate!()
(with no params) [@esemeniuc] -
[#755] Add
kebab-case
torename_all
[@iamsiddhant05] -
[#735] Support
rustls
[@jplatte]Adds
-native-tls
or-rustls
on each runtime feature:# previous features = [ "runtime-async-std" ] # now features = [ "runtime-async-std-native-tls" ]
-
[#718] Support tuple structs with
#[derive(FromRow)]
[@dvermd]
-
[#784] Use
futures_channel::oneshot
in worker for big perf win [@markazmierczak]
-
[#745] Always prefer parsing of the non-localized notice severity field [@dstoeckel]
-
[#743] Consider
utf8mb4_bin
as a string [[@digorithm]] -
[#739] Fix minor protocol detail with
iteration-count
that was blocking Vitess [@mcronce]
-
Enable compile-time type checking from cached metadata to enable building in an environment without access to a development database (e.g., Docker, CI).
-
Initial support for Microsoft SQL Server. If there is something missing that you need, open an issue. We are happy to help.
-
SQL migrations, both with a CLI tool and programmatically loading migrations at runtime.
-
Runtime-determined database driver,
Any
, to support compile-once and run with a database driver selected at runtime. -
Support for user-defined types and more generally overriding the inferred Rust type from SQL with compile-time SQL verification.
- [#418] Support zero dates and times [@blackwolf12333]
-
[#174] Inroduce a builder to construct connections to bypass the URL parsing
// MSSQL let conn = MssqlConnectOptions::new() .host("localhost") .database("master") .username("sa") .password("Password") .connect().await?; // SQLite let conn = SqliteConnectOptions::from_str("sqlite://a.db")? .foreign_keys(false) .connect().await?;
-
[#127] Get the last ID or Row ID inserted for MySQL or SQLite
// MySQL let id: u64 = query!("INSERT INTO table ( col ) VALUES ( ? )", val) .execute(&mut conn).await? .last_insert_id(); // LAST_INSERT_ID() // SQLite let id: i64 = query!("INSERT INTO table ( col ) VALUES ( ?1 )", val) .execute(&mut conn).await? .last_insert_rowid(); // sqlite3_last_insert_rowid()
-
[#263] Add hooks to the Pool:
after_connect
,before_release
, andafter_acquire
// PostgreSQL let pool = PgPoolOptions::new() .after_connect(|conn| Box::pin(async move { conn.execute("SET application_name = 'your_app';").await?; conn.execute("SET search_path = 'my_schema';").await?; Ok(()) })) .connect("postgres:// …").await?
-
[#308] [#495] Extend
derive(FromRow)
with support for#[sqlx(default)]
on fields to allow reading in a partial query [@OriolMunoz] -
[#454] [[#456]] Support
rust_decimal::Decimal
as an alternative tobigdecimal::BigDecimal
forNUMERIC
columns in MySQL and PostgreSQL [@pimeys] -
[#181] Column names and type information is now accessible from
Row
viaRow::columns()
orRow::column(name)
- [#197] [#271] Add initial support for
INTERVAL
(full support pending atime::Period
type) [@dimtion]
-
Types are now inferred for expressions. This means its now possible to use
query!
andquery_as!
for:let row = query!("SELECT 10 as _1, x + 5 as _2 FROM table").fetch_one(&mut conn).await?; assert_eq!(row._1, 10); assert_eq!(row._2, 5); // 5 + x?
-
[#167] Support
foreign_keys
explicitly with aforeign_keys(true)
method available onSqliteConnectOptions
which is a builder for new SQLite connections (and can be passed intoPoolOptions
to build a pool).let conn = SqliteConnectOptions::new() .foreign_keys(true) // on by default .connect().await?;
-
[#430] [#438] Add method to get the raw SQLite connection handle [@agentsim]
// conn is `SqliteConnection` // this is not unsafe, but what you do with the handle will be let ptr: *mut libsqlite3::sqlite3 = conn.as_raw_handle();
-
[#164] Support
TIMESTAMP
,DATETIME
,DATE
, andTIME
viachrono
in SQLite [@felipesere] [@meteficha]
-
Transaction
now mutably borrows a connection instead of owning it. This enables a new (or nested) transaction to be started from&mut conn
. -
[#145] [#444] Use a least-recently-used (LRU) cache to limit the growth of the prepared statement cache for SQLite, MySQL, and PostgreSQL [@pimeys]
- [#499]
INTEGER
now resolves toi64
instead ofi32
,INT4
will still resolve toi32
-
[#281] Deallocate SQLite statements before closing the SQLite connection [@hasali19]
-
[#284] Fix handling of
0
forBigDecimal
in PostgreSQL and MySQL [@abonander]
-
[#256] Add
query_unchecked!
andquery_file_unchecked!
with similar semantics toquery_as_unchecked!
[@meh] -
[#252] [#297] Derive several traits for the
Json<T>
wrapper type [@meh] -
[#261] Add support for
#[sqlx(rename_all = "snake_case")]
to#[derive(Type)]
[@shssoichiro] -
[#253] Add support for UNIX domain sockets to PostgreSQL [@Nilix007]
-
[#251] Add support for textual JSON on MySQL [@blackwolf12333]
-
[#275] [#268] Optionally log formatted SQL queries on execution [@shssoichiro]
-
[#267] Support Cargo.toml relative
.env
files; allows for each crate in a workspace to use their own.env
file and thus their ownDATABASE_URL
[@xyzd]
-
[#241] Type name for custom enum is not always attached to TypeInfo in PostgreSQL
-
[#237] [#238] User-defined type name matching is now case-insensitive in PostgreSQL [@qtbeee]
-
[#231] Handle empty queries (and those with comments) in SQLite
-
[#228] Provide
MapRow
implementations for functions (enables.map(|row| ...)
over.try_map(|row| ...)
)
-
[#234] Add support for
NUMERIC
in MySQL with thebigdecimal
crate [@xiaopengli89] -
[#227] Support
#[sqlx(rename = "new_name")]
on struct fields within aFromRow
derive [@sidred]
- [#214] Handle percent-encoded usernames in a database URL [@jamwaffles]
-
[#216] Mark
Cursor
,Query
,QueryAs
,query::Map
, andTransaction
as#[must_use]
[@Ace4896] -
[#213] Remove matches dependency and use matches macro from std [@nrjais]
- [#212] Removed sneaky
println!
inMySqlCursor
-
[#203] Allow an empty password for MySQL
-
[#204] Regression in error reporting for invalid SQL statements on PostgreSQL
-
[#200] Fixes the incorrect handling of raw (
r#...
) fields of a struct in theFromRow
derive [@sidred]
-
sqlx::Row
now has a lifetime ('c
) tied to the database connection. In effect, this means that you cannot storeRow
s or collect them into a collection.Query
(returned fromsqlx::query()
) hasmap()
which takes a function to map from theRow
to another type to make this transition easier.In 0.2.x
let rows = sqlx::query("SELECT 1") .fetch_all(&mut conn).await?;
In 0.3.x
let values: Vec<i32> = sqlx::query("SELECT 1") .map(|row: PgRow| row.get(0)) .fetch_all(&mut conn).await?;
To assist with the above,
sqlx::query_as()
now supports querying directly into tuples (up to 9 elements) or struct types with a#[derive(FromRow)]
.// This extension trait is needed until a rust bug is fixed use sqlx::postgres::PgQueryAs; let values: Vec<(i32, bool)> = sqlx::query_as("SELECT 1, false") .fetch_all(&mut conn).await?;
-
HasSqlType<T>: Database
is nowT: Type<Database>
to mirrorEncode
andDecode
-
Query::fetch
(returned fromquery()
) now returns a newCursor
type.Cursor
is a Stream-like type where the item type borrows into the stream (which itself borrows from connection). This means that usingquery().fetch()
you can now stream directly from the database with zero-copy and zero-allocation. -
Remove
PgTypeInfo::with_oid
and replace withPgTypeInfo::with_name
-
Results from the database are now zero-copy and no allocation beyond a shared read buffer for the TCP stream ( in other words, almost no per-query allocation ). Bind arguments still do allocate a buffer per query.
-
[#129] Add support for SQLite. Generated code should be very close to normal use of the C API.
- Adds
Sqlite
,SqliteConnection
,SqlitePool
, and other supporting types
- Adds
-
[#97] [#134] Add support for user-defined types. [@Freax13]
-
Rust-only domain types or transparent wrappers around SQL types. These may be used transparently inplace of the SQL type.
#[derive(sqlx::Type)] #[repr(transparent)] struct Meters(i32);
-
Enumerations may be defined in Rust and can match SQL by integer discriminant or variant name.
#[derive(sqlx::Type)] #[repr(i32)] // Expects a INT in SQL enum Color { Red = 1, Green = 2, Blue = 3 }
#[derive(sqlx::Type)] #[sqlx(rename = "TEXT")] // May also be the name of a user defined enum type #[sqlx(rename_all = "lowercase")] // similar to serde rename_all enum Color { Red, Green, Blue } // expects 'red', 'green', or 'blue'
-
Postgres further supports user-defined composite types.
#[derive(sqlx::Type)] #[sqlx(rename = "interface_type")] struct InterfaceType { name: String, supplier_id: i32, price: f64 }
-
-
[#98] [#131] Add support for asynchronous notifications in Postgres (
LISTEN
/NOTIFY
). [@thedodd]-
Supports automatic reconnection on connection failure.
-
PgListener
implementsExecutor
and may be used to execute queries. Be careful however as if the intent is to handle and process messages rapidly you don't want to be tying up the connection for too long. Messages received during queries are buffered and will be delivered on the next call torecv()
.
let mut listener = PgListener::new(DATABASE_URL).await?; listener.listen("topic").await?; loop { let message = listener.recv().await?; println!("payload = {}", message.payload); }
-
-
Add unchecked variants of the query macros. These will still verify the SQL for syntactic and semantic correctness with the current database but they will not check the input or output types.
This is intended as a temporary solution until
query_as!
is able to support user defined types.query_as_unchecked!
query_file_as_unchecked!
-
Add support for many more types in Postgres
JSON
,JSONB
[@oeb25]INET
,CIDR
[@PoiScript]- Arrays [@oeb25]
- Composites ( Rust tuples or structs with a
#[derive(Type)]
) NUMERIC
[@abonander]OID
(u32
)"CHAR"
(i8
)TIMESTAMP
,TIMESTAMPTZ
, etc. with thetime
crate [@utter-step]- Enumerations ( Rust enums with a
#[derive(Type)]
) [@Freax13]
-
Query
(andQueryAs
; returned fromquery()
,query_as()
,query!()
, andquery_as!()
) now will accept both&mut Connection
or&Pool
where as in 0.2.x they required&mut &Pool
. -
Executor
now takes any value that implementsExecute
as a query.Execute
is implemented forQuery
andQueryAs
to mean exactly what they've meant so far, a prepared SQL query. However,Execute
is also implemented for just&str
which now performs a raw or unprepared SQL query. You can further use this to fetchRow
s from the database though it is not as efficient as the prepared API (notably Postgres and MySQL send data back in TEXT mode as opposed to in BINARY mode).use sqlx::Executor; // Set the time zone parameter conn.execute("SET TIME ZONE LOCAL;").await // Demonstrate two queries at once with the raw API let mut cursor = conn.fetch("SELECT 1; SELECT 2"); let row = cursor.next().await?.unwrap(); let value: i32 = row.get(0); // 1 let row = cursor.next().await?.unwrap(); let value: i32 = row.get(0); // 2
-
Query
(returned fromquery()
) no longer hasfetch_one
,fetch_optional
, orfetch_all
. You must map the row usingmap()
and then you will have aquery::Map
value that has the former methods available.let values: Vec<i32> = sqlx::query("SELECT 1") .map(|row: PgRow| row.get(0)) .fetch_all(&mut conn).await?;
-
[#62] [#130] [#135] Remove explicit set of
IntervalStyle
. Allow usage of SQLx for CockroachDB and potentially PgBouncer. [@bmisiak] -
[#108] Allow nullable and borrowed values to be used as arguments in
query!
andquery_as!
. For example, where the column would resolve toString
in Rust (TEXT, VARCHAR, etc.), you may now useOption<String>
,Option<&str>
, or&str
instead. [@abonander] -
[#108] Make unknown type errors far more informative. As an example, trying to
SELECT
aDATE
column will now try and tell you about thechrono
feature. [@abonander]optional feature `chrono` required for type DATE of column #1 ("now")
-
[#125] [#126] Fix statement execution in MySQL if it contains NULL statement values [@repnop]
-
[#105] [#109] Allow trailing commas in query macros [@timmythetiny]
-
Fix decoding of Rows containing NULLs in Postgres #104
-
After a large review and some battle testing by @ianthetechie of the
Pool
, a live leaking issue was found. This has now been fixed by @abonander in #84 which included refactoring to make the pool internals less brittle (using RAII instead of manual work is one example) and to help any future contributors when changing the pool internals. -
Passwords are now being percent-decoded before being presented to the server [@repnop]
-
[@100] Fix
FLOAT
andDOUBLE
decoding in MySQL
-
[#72] Add
PgTypeInfo::with_oid
to allow simple construction ofPgTypeInfo
which enablesHasSqlType
to be implemented by downstream consumers of SQLx [@jplatte] -
[#96] Add support for returning columns from
query!
with a name of a rust keyword by using raw identifiers [@yaahc] -
[#71] Implement derives for
Encode
andDecode
. This is the first step to supporting custom types in SQLx. [@Freax13]
- Fix decoding of Rows containing NULLs in MySQL (and add an integration test so this doesn't break again)
- Fix
query!
when used on a query that does not return results
-
Fix stall when requesting TLS from a Postgres server that explicitly does not support TLS (such as postgres running inside docker) [@abonander]
-
[#66] Declare used features for
tokio
insqlx-macros
explicitly
- [#64, #65] Fix decoding of Rows containing NULLs in MySQL [@danielakhterov]
- [#55] Use a shared tokio runtime for the
query!
macro compile-time execution (under theruntime-tokio
feature) [@udoprog]
-
Support Tokio through an optional
runtime-tokio
feature. -
Support SQL transactions. You may now use the
begin()
function onPool
orConnection
to start a new SQL transaction. This returnssqlx::Transaction
which willROLLBACK
onDrop
or can be explicitlyCOMMIT
usingcommit()
. -
Support TLS connections.
-
Support for
SCRAM-SHA-256
authentication in Postgres #37 @danielakhterov -
Implement
Debug
for Pool #42 @prettynatty
-
Support for Authentication in MySQL 5+ including the newer authentication schemes now default in MySQL 8:
mysql_native_password
,sha256_password
, andcaching_sha2_password
. -
Chrono
support for MySQL was only partially implemented (was missingNaiveTime
andDateTime<Utc>
). -
Vec<u8>
(and[u8]
) support for MySQL (BLOB
) and Postgres (BYTEA
).