<!--- Provide a general summary of the issue in the Title above -->

I am creating a many 2 many table with one primary key as bigserial auto-increment and then the two FK with a unique group. The two foreign keys ids are UUIDs.

The tables look like that:

type UserTest struct {
	ID          uuid.UUID        `pg:",type:uuid"`
	Permissions []PermissionTest `pg:"many2many:upt"`
}

type PermissionTest struct {
	ID uuid.UUID `pg:",type:uuid"`
}

type UsersPermissionsTest struct {
	tableName struct{} `pg:"upt"`

	ID int `pg:"id,pk"`

	PermissionID uuid.UUID `pg:"permission_test_id,type:uuid,unique:idx_upt_unique"`
	UserID       uuid.UUID `pg:"user_test_id,type:uuid,unique:idx_users_upt_unique"`
}

Current Behavior

When fetching the relation of the many 2 many table instead of setting the permission id to the corresponding column in the select the ORM is trying to set the pk which is int to the FK permission id which is a UUID. (IDK if it's clear)

This is the final selection:

SELECT "users_permissions_test".*, "permission_test"."id" FROM "permission_tests" AS "permission_test" 
JOIN "upt" AS "users_permissions_test" ON ("users_permissions_test"."user_test_id") IN ('25289c7f-e733-4903-a8a3-e1569712ea69')
WHERE ("permission_test"."id" = "users_permissions_test"."permission_test_id")

The ORM is trying to assign "user_permissions_test"."id" to the Permission.ID struct. Which lead to panic: uuid: incorrect UUID length

This error happens with the UUID type. If you replace Permissions Ids and User ids with int it works suddenly as expected. The Permission.ID is correctly set from the "user_permissions_test"."permission_test_id" and not from the "user_permissions_test"."id". (that is surprising)

Expected Behavior

The ORM must correctly set the Permission ID from the "user_permissions_test"."permission_test_id" and not from the "user_permissions_test"."id".

Possible Solution

Yesterday I was diving into your code and it seems like the many to many scanner have an empty columns map field while scanning the first column (which is the id from user_permissions_test).

Likely too deep to understand why this bug happens. I just saw today that the problem only appear with UUID I first though that it was a bug because my foreign keys aren't the primary composite key, but it's not the case.

Steps to Reproduce

Launch this code:

package main

import (
	"fmt"

	"github.com/go-pg/pg/extra/pgdebug"
	"github.com/go-pg/pg/v10"
	"github.com/go-pg/pg/v10/orm"
	uuid "github.com/satori/go.uuid"
)

type UserTest struct {
	ID          uuid.UUID        `pg:",type:uuid"`
	Permissions []PermissionTest `pg:"many2many:upt"`
}

type PermissionTest struct {
	ID uuid.UUID `pg:",type:uuid"`
}

type UsersPermissionsTest struct {
	tableName struct{} `pg:"upt"`

	ID int `pg:"id,pk"`

	PermissionID uuid.UUID `pg:"permission_test_id,type:uuid,unique:idx_upt_unique"`
	UserID       uuid.UUID `pg:"user_test_id,type:uuid,unique:idx_users_upt_unique"`
}

func Example() {
	opts, _ := pg.ParseURL("postgresql://asvf-montagne-api:asvf-montagne-api@localhost:5172/asvf-montagne-api?sslmode=disable")
	db := pg.Connect(opts)

	defer db.Close()

	db.AddQueryHook(pgdebug.DebugHook{
		Verbose: true,
	})

	if err := createManyToManyTables(db); err != nil {
		panic(err)
	}

	userid := uuid.NewV4()
	permid := uuid.NewV4()
	values := []interface{}{
		&UserTest{ID: userid},
		&PermissionTest{ID: permid},
		&UsersPermissionsTest{ID: 30, UserID: userid, PermissionID: permid},
	}
	for _, v := range values {
		_, err := db.Model(v).Insert()
		if err != nil {
			panic(err)
		}
	}

	u := new(UserTest)
	err := db.Model(u).Relation("Permissions").First()
	if err != nil {
		panic(err)
	}

	fmt.Println("User", u.ID, "Permission", u.Permissions)
}

func createManyToManyTables(db *pg.DB) error {
	orm.RegisterTable((*UsersPermissionsTest)(nil))

	models := []interface{}{
		(*UserTest)(nil),
		(*PermissionTest)(nil),
		(*UsersPermissionsTest)(nil),
	}
	for _, model := range models {
		err := db.Model(model).CreateTable(&orm.CreateTableOptions{
			Temp: true,
		})
		if err != nil {
			return err
		}
	}
	
	return nil
}

func main() {

	Example()
}

You should have this output:

CREATE TEMP TABLE "user_tests" ("id" uuid, PRIMARY KEY ("id"))
CREATE TEMP TABLE "permission_tests" ("id" uuid, PRIMARY KEY ("id"))
CREATE TEMP TABLE "upt" ("id" bigserial, "permission_test_id" uuid, "user_test_id" uuid, PRIMARY KEY ("id"), UNIQUE ("permission_test_id"), UNIQUE ("user_test_id"))
INSERT INTO "user_tests" ("id") VALUES ('25289c7f-e733-4903-a8a3-e1569712ea69')
INSERT INTO "permission_tests" ("id") VALUES ('86aca506-d00e-4028-aa22-10939dd4d326')
INSERT INTO "upt" ("id", "permission_test_id", "user_test_id") VALUES (30, '86aca506-d00e-4028-aa22-10939dd4d326', '25289c7f-e733-4903-a8a3-e1569712ea69')
SELECT "user_test"."id" FROM "user_tests" AS "user_test" ORDER BY "user_test"."id" LIMIT 1
SELECT "users_permissions_test".*, "permission_test"."id" FROM "permission_tests" AS "permission_test" JOIN "upt" AS "users_permissions_test" ON ("users_permissions_test"."user_test_id") IN ('25289c7f-e733-4903-a8a3-e1569712ea69') WHERE ("permission_test"."id" = "users_permissions_test"."permission_test_id")
panic: uuid: incorrect UUID length: 30 <--- the error

Context (Environment)

I'm just trying to bind what I have in my database. The link table has fields with the id as a primary key and two other FK as UUIDs.

There is no black magic it's a simple case, I don't know why it isn't working, I was expecting it to works.

try to specify conflict fields: .OnConflict("(id) DO NOTHING")

Should I update the tests to use a different way to looking at private properties? If the goal is to verify the structure of generated files, maybe we could use reflection and assert on that.

assertAttributeSame() is deprecated and will be removed in PHPUnit 9.
readAttribute() is deprecated and will be removed in PHPUnit 9.
getObjectAttribute() is deprecated and will be removed in PHPUnit 9.
assertAttributeInstanceOf() is deprecated and will be removed in PHPUnit 9.

As v4 doesn't support PHP 8.

Replaced with https://github.com/tebru/gson-php/pull/55

I've used the following and it's worked, but idk if it's the intended way of using it:

db.Relation("User.id").Relation("User.name")

I agree though that it'd be simpler if it were just a simple Relation call.

This piggybacks off the discussion around https://github.com/hashicorp/raft/pull/124 While we acknowledge this discrepancy in timeout assumptions, it could still be valuable to at least expose the option to buffer the applyCh

This piggybacks off the discussion around https://github.com/hashicorp/raft/pull/124 While we acknowledge this discrepancy in timeout assumptions, it could still be valuable to at least expose the option to buffer the applyCh

This change implements the solution suggested in https://github.com/hashicorp/raft/pull/124 to enable batch processing off the raft applyCh

For context, our application uses this library, and we are currently bottlenecked by StoreLogs throughput. We believe we can boost this throughput by increasing batch size in StoreLogs, which fsync's on each call.

Appreciate all the help team 💯

Issue tracker is used for reporting bugs and discussing new features. Please use Discord or stackoverflow for supporting issues.

<!--- Provide a general summary of the issue in the Title above -->

Expected Behavior

Support for selecting multiple columns when using Relation.

<!--- Tell us what should happen -->

Current Behavior

Currently It just selects one column only no matter how many columns you specify.

<!--- Tell us what happens instead of the expected behavior -->

Possible Solution

db.Relation("User.id").ComunExpr(`"user"."name" as "user__name"`)

<!--- Not obligatory, but suggest a fix/reason for the bug, -->

Context (Environment)

In many cases we have to select multiple columns which is not possible without use of ColumnExpr(), supporting within a relation would greatly facilitate.

<!--- How has this issue affected you? What are you trying to accomplish? --> <!--- Providing context helps us come up with a solution that is most useful in the real world -->

<!--- Provide a general summary of the issue in the Title above -->

Detailed Description

<!--- Provide a detailed description of the change or addition you are proposing -->

Possible Implementation

db.Relation("User,id,name")

<!--- Not obligatory, but suggest an idea for implementing addition or change -->

I've the same issue while working on many2many relation.

TODO:

• [ ] fix this comment and docs generally if we are happy with this API approach (duplicating fields)

We are still having some issues with tests in CI, but the suite passes locally for me and the new tests I added pass using -race too FWIW.

These parameters are reasonably easy to make reloadable in a running raft instance.

Other parameters like the timeouts may be too although some require more thought about whether they are safe to change at runtime, however I'm sticking with these three right now as they are the critical parameters needed to get out of an otherwise unrecoverable situation detailed in https://github.com/hashicorp/consul/issues/9609.

There are other possible mitigations and solutions to that problem including my previous WIP PR that prevents snapshotting during replication entirely, but this has fewer uncertainties about safety and is the simplest change that gives a manual intervention option where almost none exists today.

If others are happy with this change, we can merge it and then allow Consul (and others) to have a way to re-configure snapshotting and log pruning on a leader without restarting it and loosing leadership.

Noticed while submitting https://github.com/hashicorp/raft/pull/442 that this repo is missing tests for the latest two versions of Go.

Thank you for your submission! We require that all contributors sign our Contributor License Agreement ("CLA") before we can accept the contribution. Read and sign the agreement

Learn more about why HashiCorp requires a CLA and what the CLA includes

_{Have you signed the CLA already but the status is still pending? Recheck it.}

I want to do something like this, apparenty it dosn't work.

var names []string
db.Query(pg.Scan(&names), `select name from foo`)

For a single column you can use

var strs pg.Strings
db.Query(&strs, `select name from foo`)