go-sql-driver / mysql

* stop rounding times

fixes #1121

* trim trailing zeros

* improve GitHub Actions workflows

- add windows to build matrix
- use newer versions of MySQL and MariaDB

* remove branch restriction

* Run macOS CI on Github Actions

* no cover

* Fix workflow name

* Actions: Add linux test

* fix

* fix

* fix go 1.10

* Remove .travis.yml

#1133)

* return sql.NullTime if it available

* NullTime should be used with parseTime=true option

* Add deprecated message to NullTime
* remove NullTime from README

https://golang.org/pkg/database/sql/driver/#Value says:

> Value is a value that drivers must be able to handle. It is either nil, a type handled by a database driver's NamedValueChecker interface, or an instance of one of these types:

* fixed the way of parsing datetime when byte slice string

The benchmark results

$ go test -benchmem . -bench "^BenchmarkParseByte"
goos: darwin
goarch: amd64
pkg: github.com/go-sql-driver/mysql
BenchmarkParseByteDateTime-4                    12023173               104 ns/op               0 B/op          0 allocs/op
BenchmarkParseByteDateTimeStringCast-4           3394239               355 ns/op              32 B/op          1 allocs/op

* added line to AUTHORS file

* fixed error handling

* fixed nanosec digits

* added more tests for error

* renamed parseByteDateTime to parseDateTime

* reverted base null time

* Update utils.go

Co-authored-by: Inada Naoki <songofacandy@gmail.com>

* Update utils.go

Co-authored-by: Inada Naoki <songofacandy@gmail.com>

* Update utils.go

Co-authored-by: Inada Naoki <songofacandy@gmail.com>

* removed deprecatedParseDateTime from test

Co-authored-by: Inada Naoki <songofacandy@gmail.com>

* Replace whitelist/blacklist terminology with allowlist/denylist

* Add myself to AUTHORS

* PR feedback

* Denylist --> denied

* Update denied --> rejected

* connect: reduce allocations when building SET command

* handleParams: use strings.Builder instead of direct []byte

* travis: update image versions

* travis: use mysql client inside the docker container for polling

* travis: force mysql client to connect via tcp

* travis: use root pass in mysql client

* travis: use empty root password

The fuzzer targets the `Open` function, which can be run locally as well as through oss-fuzz's infrastructure.

When opening a connection, instead of iterating on all variables and
calling "SET <variable>=<value>" for each, we now use a single SET statement
with all pairs using a comma as separator:
  SET <variable1>=<value1>,<variable2>=<value2>,...

* Travis-CI: reverse order of Go version

Reverse order of Go versions so the recent ones are tested first.

* Travis-CI: add Go 1.14

* Travis-CI: move 'tip' just below the latest Go release

* Travis-CI: upgrade Go to 1.14 to run tests

If
    * BeginTx is called with a non-default isolation level,
    * The context is canceled before SET TRANSACTION ISOLATION LEVEL
    completes,
then the connection:
    * has the cancelled property set to "context cancelled",
    * has the closed property set to true,
and,
    * BeginTx returns "context canceled"

Because of this, the connection gets put back into the connection pool.
When it is checked out again, if BeginTx is called on it again _without_
an isolation level,
    * then we fall into the mc.closed.IsSet() check in begin(),
    * so we return ErrBadConn,
    * so the driver kicks the broken connection out of the pool
    * (and transparently retries to get a new connection that isn't
    broken too).

However, if BeginTx is called on the connection _with_ an isolation
level, then we return a context canceled error from the SET TRANSACTION
ISOLATION LEVEL call.

That means the broken connection will stick around in the pool forever
(or until it's checked out for some other operation that correctly
returns ErrBadConn).

The fix is to check for the connection being closed before executing SET
TRANSACTION ISOLATION LEVEL.

* Add CheckNamedValue for mysqlStmt

* Update AUTHORS

Co-authored-by: Zhixin Wen <zwen@nuro.ai>

* travis: allow master branch to fail

* travis: matrix is an alias for jobs

* travis: remove obsolete sudo key

* travis: remove obsolete sudo keys in jobs matrix

Implements a Travis CI task that checks if the driver compiles on all platforms supported by Go.

Fixes #1050

According to the linked documentation at
http://dev.mysql.com/doc/internals/en/connection-phase-packets.html#packet-Protocol::SSLRequest
SSLRequest packet should have zero filler similar to the regular handshake request,
but now the driver puts zeros only in the regular request. Luckily vanilla MySQL
doesn't rely on this zero filler and doesn't verify its presence, thus the driver worked
fine so far. But MySQL can change to rely on zeros at any point.

The problem was discovered while testing against a customized MySQL.

Following #1058, in order for the driver.Value to get as a json.RawMessage,
the converter should accept it as a valid value, and handle it as bytes in
case where interpolation is disabled

json encoded data is represented as bytes however it should be interpolated as a string

Fixes #819

* CHANGELOG: include v1.4.1

* Release v1.5.0

* conncheck: allow to disable via config

* dsn: refactor writing of params in FormatDSN