The new logic checks for any type of nil at the beginning of Encode and
then either treats it as NULL or calls the driver.Valuer method if
appropriate.
This should preserve the existing nil normalization while restoring the
ability to encode nil driver.Valuer values.
Extract logic for finding OID and converting argument to encodable
value. This is in preparation for a future change for better supporting
nil driver.Valuer values.
Future commit will be using bytes.Clone which was implemented in Go
1.20.
Also update README.md to reflect that minimum supported Go version is
1.21. But only requiring Go 1.20 in go.mod to avoid needlessly breaking
old Go when it still works.
pgx v5 introduced nil normalization for typed nils. This means that
[]byte(nil) is normalized to nil at the edge of the encoding system.
This simplified encoding logic as nil could be encoded as NULL and type
specific handling was unneeded.
However, database/sql compatibility requires Value to be called on a
nil pointer that implements driver.Valuer. This was broken by
normalizing to nil.
This commit changes the normalization logic to not normalize pointers
that directly implement driver.Valuer to nil. It still normalizes
pointers that implement driver.Valuer through implicit derefence.
e.g.
type T struct{}
func (t *T) Value() (driver.Value, error) {
return nil, nil
}
type S struct{}
func (s S) Value() (driver.Value, error) {
return nil, nil
}
(*T)(nil) will not be normalized to nil but (*S)(nil) will be.
https://github.com/jackc/pgx/issues/1566
A single Connect("connstring") may actually make multiple connection
requests due to TLS or HA configuration. Previously, when all attempts
failed only the last error was returned. This could be confusing.
Now details of all failed attempts are included.
For example, the following connection string:
host=localhost,127.0.0.1,foo.invalid port=1,2,3
Will now return an error like the following:
failed to connect to `user=postgres database=pgx_test`:
lookup foo.invalid: no such host
[::1]:1 (localhost): dial error: dial tcp [::1]:1: connect: connection refused
127.0.0.1:1 (localhost): dial error: dial tcp 127.0.0.1:1: connect: connection refused
127.0.0.1:2 (127.0.0.1): dial error: dial tcp 127.0.0.1:2: connect: connection refused
https://github.com/jackc/pgx/issues/1929
Existing generic helpers always call defer rows.Close().
Examples of their usage also omit external defer rows.Close() call.
For clarity, state that explicitly, because that's another point
why one would want to switch to generic helpers from manually written
rows.Next() loop.
If ScanLocation is set, the timestamps will be assumed to be in the
given location when scanning from the database.
The Codec interface is now implemented by *pgtype.TimestampCodec instead
of pgtype.TimestampCodec. This is technically a breaking change, but it
is extremely unlikely that anyone is depending on this, and if there is
downstream breakage it is trivial to fix.
https://github.com/jackc/pgx/issues/1195https://github.com/jackc/pgx/issues/1945
If ScanLocation is set, it will be used to convert the time to the given
location when scanning from the database.
The Codec interface is now implemented by *pgtype.TimestamptzCodec
instead of pgtype.TimestamptzCodec. This is technically a breaking
change, but it is extremely unlikely that anyone is depending on this,
and if there is downstream breakage it is trivial to fix.
https://github.com/jackc/pgx/issues/1195https://github.com/jackc/pgx/issues/1945
Modify the RowToStructByPos/Name functions to store the computed mapping
of columns to struct field locations in a cache to reuse between calls.
Because this computation can be expensive and the same few results will
frequently be reused, caching these results provides a significant
speedup.
For positional mappings, we can key the cache by just the struct-type.
However, for named mappings, the key must include a representation of
the columns, in order, since different columns produce different
mappings.
It is more often that interesting information is buried by the verbose
output than the verbose output is useful. It can be reenabled later if
necessary.
Tests were failing with:
Error: Process completed with exit code 143.
This appears to mean that Github Actions killed the runner.
See https://github.com/jackc/pgx/actions/runs/8216337993/job/22470808811
for an example.
It appears Github Actions kills runners based on resource usage. Running
tests one at a time reduces the resource usage and avoids the problem.
Or at least that's what I presume is happening. It sure is fun debugging
issues on cloud systems where you have limited visibility... :(
fixes https://github.com/jackc/pgx/issues/1934
Jack Christensen
ngavinsir
Oleksandr Redko
Mitar
Pavlo Golub
Carlos Pérez-Aradros Herce
WGH
Oleksandr Redko
Zach Olstein
Felix Röhrich
Tomas Zahradnicek