Documentation
¶
Overview ¶
Package store is the typed data access of a model: CRUD with optimistic locking and soft delete, filtered and paged lists, aggregates, and an owner scope that fences every query to the caller's rows.
posts := store.New[Post](d)
p, err := posts.Get(ctx, store.RID(id)) // 404 when missing or not yours
page, err := posts.List(ctx, where.Eq("status", "draft"), where.Page(1, 20))
err = posts.Update(ctx, store.RID(id), store.Patch(req).Onto(p))
Fields ¶
A field is named by its public name: its JSON name, or its column when the model's JSON has no name for it (json:"-", or a field inside a named struct). id is the RID; the numeric key has no name (store.ID, PluckIDs, ListByIDs).
The store tags say what a client may touch: store:"query" is what a URL filters and orders on (ListFromQuery and chok.List, QueryFields, the OpenAPI filters), store:"update" what Patch writes from a request. id, created_at and updated_at are always queryable. A model with no store tag at all is refused by New: the whitelist is never implicit.
The code is not fenced by the tags: where options, store.Where, aggregates, Pluck, ListIn, Upsert's conflict fields and Set take every column of the model, the managed ones too (Set and Patch never write those). A field name that the code got from a client must be checked against QueryFields first; an unknown name in code is a programming error, a 500 that still matches where.ErrUnknownField. Alias maps a field to another column.
Owner scope ¶
A model embedding db.OwnedModel or db.OwnedSoftDeleteModel is fenced to chok.PrincipalFrom(ctx): reads see only the principal's rows (none without a principal), writes stamp and check the owner, and a write without a principal fails with NoPrincipalError. Principals holding one of the policy's admin roles see every row.
Errors ¶
Every error a handler should answer with implements chok.Coder: NotFoundError (404), VersionConflictError and DuplicateError (409), EmptyChangesError (400), NoPrincipalError (401). Return them as they are.
Example ¶
package main
import (
"context"
"errors"
"fmt"
"os"
"path/filepath"
"github.com/zynthara/chok"
"github.com/zynthara/chok/db"
"github.com/zynthara/chok/store"
"github.com/zynthara/chok/store/where"
)
// Article is an owned model: store tags declare what the API may filter
// on and update.
type Article struct {
db.OwnedSoftDeleteModel
Title string `json:"title" store:"query,update" gorm:"size:200;not null"`
Status string `json:"status" store:"query,update" gorm:"size:20;not null"`
}
func (Article) RIDPrefix() string { return "art" }
// UpdateArticle is a PATCH body: a nil field was not sent.
type UpdateArticle struct {
ID string `path:"id"`
Version *int `json:"version"`
Title *string `json:"title"`
Status *string `json:"status"`
}
func openExample() (*db.DB, func()) {
dir, _ := os.MkdirTemp("", "chok-store-example")
d := db.MustOpen(db.Config{Driver: "sqlite", SQLite: db.SQLiteConfig{Path: filepath.Join(dir, "app.db")}},
db.Tables(db.Table(&Article{}, db.SoftUnique("uk_article_title_owner", "title", "owner_id"))))
return d, func() { _ = d.Close(context.Background()); _ = os.RemoveAll(dir) }
}
func main() {
d, done := openExample()
defer done()
articles := store.New[Article](d)
alice := chok.WithPrincipal(context.Background(), chok.Principal{UserID: "alice"})
a := &Article{Title: "Hello", Status: "draft"}
if err := articles.Create(alice, a); err != nil {
fmt.Println(err)
return
}
fmt.Println(a.OwnerID, a.Version)
// A partial update locked on the version the client saw.
title := "Hello, store"
req := &UpdateArticle{ID: a.RID, Version: new(1), Title: &title}
if err := articles.Update(alice, store.RID(req.ID), store.Patch(req).Onto(a)); err != nil {
fmt.Println(err)
return
}
fmt.Println(a.Title, a.Version)
// Replaying the stale version is a 409.
err := articles.Update(alice, store.RID(req.ID), store.Patch(req).Onto(a))
fmt.Println(errors.Is(err, chok.ErrConflict))
// Bob's queries never see alice's rows.
bob := chok.WithPrincipal(context.Background(), chok.Principal{UserID: "bob"})
_, err = articles.Get(bob, store.RID(a.RID))
fmt.Println(errors.Is(err, chok.ErrNotFound))
page, _ := articles.List(alice, where.Eq("status", "draft"), where.Order("created_at", true), where.Count())
fmt.Println(page.Total, page.Items[0].Title)
}
Output: alice 1 Hello, store 2 true true 1 Hello, store
Index ¶
- type AggValue
- type Aggregate
- type Changes
- type DuplicateError
- type EmptyChangesError
- type Group
- type HookSet
- type Locator
- type MissingConditionsError
- type NoPrincipalError
- type NotFoundError
- type Number
- type Option
- type PatchBuilder
- type Scalar
- type ScopeFunc
- type Store
- func (s *Store[T]) Avg(ctx context.Context, field string, opts ...where.Option) (float64, bool, error)
- func (s *Store[T]) BatchCreate(ctx context.Context, objs []*T) error
- func (s *Store[T]) BatchUpsert(ctx context.Context, objs []*T, conflict ...string) error
- func (s *Store[T]) Count(ctx context.Context, opts ...where.Option) (int64, error)
- func (s *Store[T]) CountDistinct(ctx context.Context, field string, opts ...where.Option) (int64, error)
- func (s *Store[T]) Create(ctx context.Context, obj *T) error
- func (s *Store[T]) Delete(ctx context.Context, loc Locator, opts ...where.Option) error
- func (s *Store[T]) Exists(ctx context.Context, loc Locator) (bool, error)
- func (s *Store[T]) Get(ctx context.Context, loc Locator, opts ...where.Option) (*T, error)
- func (s *Store[T]) GetForUpdate(ctx context.Context, loc Locator, opts ...where.Option) (*T, error)
- func (s *Store[T]) GroupBy[K comparable](ctx context.Context, field string, aggs []Aggregate, opts ...where.Option) ([]Group[K], error)
- func (s *Store[T]) List(ctx context.Context, opts ...where.Option) (*chok.Page[T], error)
- func (s *Store[T]) ListByIDs(ctx context.Context, ids []uint, opts ...where.Option) ([]T, error)
- func (s *Store[T]) ListFromQuery(ctx context.Context, q url.Values) (*chok.Page[T], error)
- func (s *Store[T]) ListIn[V comparable](ctx context.Context, field string, values []V, opts ...where.Option) ([]T, error)
- func (s *Store[T]) ListWithCursor(ctx context.Context, opts ...where.Option) (*chok.Page[T], error)
- func (s *Store[T]) Max[N Scalar](ctx context.Context, field string, opts ...where.Option) (N, bool, error)
- func (s *Store[T]) Min[N Scalar](ctx context.Context, field string, opts ...where.Option) (N, bool, error)
- func (s *Store[T]) Pluck[F any](ctx context.Context, field string, opts ...where.Option) ([]F, error)
- func (s *Store[T]) PluckIDs(ctx context.Context, opts ...where.Option) ([]uint, error)
- func (s *Store[T]) QueryFields() []string
- func (s *Store[T]) Restore(ctx context.Context, loc Locator) error
- func (s *Store[T]) Sum[N Number](ctx context.Context, field string, opts ...where.Option) (N, bool, error)
- func (s *Store[T]) Tx(ctx context.Context, fn func(txCtx context.Context) error) error
- func (s *Store[T]) Unsafe(ctx context.Context) *gorm.DB
- func (s *Store[T]) Update(ctx context.Context, loc Locator, ch *Changes, opts ...where.Option) error
- func (s *Store[T]) Upsert(ctx context.Context, obj *T, conflict ...string) error
- type VersionConflictError
Examples ¶
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
This section is empty.
Types ¶
type AggValue ¶ added in v0.2.0
type AggValue struct {
// contains filtered or unexported fields
}
AggValue is one aggregate of a Group. Its kind follows from the Aggregate: counts and integer sum, min and max are Int64, avg and float columns Float64, min and max of a time column Time. A group whose values were all NULL is IsNull. It encodes to JSON as a number, a time or null.
func (AggValue) Float64 ¶ added in v0.2.0
Float64 returns a numeric aggregate, converting an integer one.
func (AggValue) Int64 ¶ added in v0.2.0
Int64 returns an integer aggregate; false for other kinds, which an int64 would truncate.
func (AggValue) MarshalJSON ¶ added in v0.2.0
MarshalJSON encodes the value as a number, an RFC 3339 time or null.
type Aggregate ¶ added in v0.2.0
Aggregate is one value GroupBy computes per group. Func is count (the group's rows; no Field), count_distinct, sum, avg, min or max of Field.
type Changes ¶
type Changes struct {
// contains filtered or unexported fields
}
Changes is the set of columns an Update writes: from Patch(...).Onto, or from Set.
func Set ¶
Set writes the given fields by public name: any column of the model but the managed ones (id, version, the timestamps, the owner, the soft-delete columns). It locks on no version unless the Update adds where.Version. It is the code's own write; a request DTO goes through Patch, which writes only store:"update" fields.
type DuplicateError ¶
type DuplicateError struct {
Field string
Constraint string
// contains filtered or unexported fields
}
DuplicateError is a write that a unique constraint refused. Field is the field the constraint guards, when known; Constraint its name.
func (*DuplicateError) APIError ¶ added in v0.2.0
func (e *DuplicateError) APIError() *chok.Error
APIError is the 409, with metadata.field.
func (*DuplicateError) Error ¶ added in v0.2.0
func (e *DuplicateError) Error() string
func (*DuplicateError) Is ¶ added in v0.2.0
func (e *DuplicateError) Is(target error) bool
Is matches chok.ErrConflict.
func (*DuplicateError) Unwrap ¶ added in v0.2.0
func (e *DuplicateError) Unwrap() error
type EmptyChangesError ¶ added in v0.2.0
type EmptyChangesError struct{}
EmptyChangesError is an update that changes nothing.
func (*EmptyChangesError) APIError ¶ added in v0.2.0
func (e *EmptyChangesError) APIError() *chok.Error
APIError is the 400.
func (*EmptyChangesError) Error ¶ added in v0.2.0
func (e *EmptyChangesError) Error() string
func (*EmptyChangesError) Is ¶ added in v0.2.0
func (e *EmptyChangesError) Is(target error) bool
Is matches chok.ErrBadRequest.
type Group ¶ added in v0.2.0
type Group[K comparable] struct { Key K `json:"key"` Values []AggValue `json:"values"` }
Group is one bucket of GroupBy: the key and the aggregates, in the order they were asked for.
type HookSet ¶ added in v0.2.0
type HookSet[T any] struct { BeforeCreate func(ctx context.Context, obj *T) error BeforeUpdate func(ctx context.Context, loc Locator, changes map[string]any) error BeforeDelete func(ctx context.Context, loc Locator) error }
HookSet holds the functions a store calls before it writes; a hook's error aborts the write. BeforeCreate may still change the object; it runs after the owner is stamped. BeforeUpdate receives a copy of the changes, field name to the value about to be written, after they passed validation. Upsert and BatchUpsert run BeforeCreate.
type Locator ¶
type Locator interface {
String() string
// contains filtered or unexported methods
}
Locator picks the row, or rows, an operation targets.
type MissingConditionsError ¶ added in v0.2.0
type MissingConditionsError struct{}
MissingConditionsError is store.Where with no filter: a programming error that would otherwise update or delete every row.
func (*MissingConditionsError) APIError ¶ added in v0.2.0
func (e *MissingConditionsError) APIError() *chok.Error
APIError is a 500: the request is fine, the code is not.
func (*MissingConditionsError) Error ¶ added in v0.2.0
func (e *MissingConditionsError) Error() string
type NoPrincipalError ¶ added in v0.2.0
type NoPrincipalError struct{}
NoPrincipalError is a write to an owner-scoped model without a principal in the context.
func (*NoPrincipalError) APIError ¶ added in v0.2.0
func (e *NoPrincipalError) APIError() *chok.Error
APIError is the 401.
func (*NoPrincipalError) Error ¶ added in v0.2.0
func (e *NoPrincipalError) Error() string
func (*NoPrincipalError) Is ¶ added in v0.2.0
func (e *NoPrincipalError) Is(target error) bool
Is matches chok.ErrUnauthorized.
type NotFoundError ¶
type NotFoundError struct {
Locator string
}
NotFoundError is a locator that matched no row the caller may see.
func (*NotFoundError) APIError ¶ added in v0.2.0
func (e *NotFoundError) APIError() *chok.Error
APIError is the 404.
func (*NotFoundError) Error ¶
func (e *NotFoundError) Error() string
func (*NotFoundError) Is ¶
func (e *NotFoundError) Is(target error) bool
Is matches chok.ErrNotFound.
type Number ¶ added in v0.2.0
Number is what Sum reads: int64 from an integer column, or float64 from any numeric column, at float precision.
type Option ¶ added in v0.2.0
type Option func(*config)
Option configures New.
func Alias ¶ added in v0.2.0
Alias maps a declared field to another column, for a JSON name that differs from the column gorm derives.
func Constraints ¶ added in v0.2.0
Constraints names the field a unique constraint or index guards, so a duplicate-key error says which field collided. The model's own unique indexes and those its db.Table declares, Unique and SoftUnique, are known without it.
func Policy ¶ added in v0.2.0
func Policy(p db.StorePolicy) Option
Policy replaces db.store, the handle's default posture, for this store. The policy is copied: changing p afterwards changes nothing.
type PatchBuilder ¶ added in v0.2.0
type PatchBuilder struct {
// contains filtered or unexported fields
}
PatchBuilder is a partial update read from a request: Onto turns it into Changes.
func Patch ¶ added in v0.2.0
func Patch(req any) *PatchBuilder
Patch reads a request DTO for a partial update. Its non-nil pointer fields that come from the JSON body are the changes, under their JSON names; a nil pointer is a field the client did not send. Fields bound from the path, the query or a header never take part, and neither does a field tagged store:"-". A field with the JSON name "version", an int or *int, is not a change: when set it is the version the client last saw, and the update fails with VersionConflictError if the row has moved on.
Patch panics if req is not a struct or a non-nil pointer to one.
func (*PatchBuilder) Onto ¶ added in v0.2.0
func (b *PatchBuilder) Onto(obj any) *Changes
Onto applies the patch to obj, the *T the handler loaded, and locks the update on obj's version (or on the request's version field). On success obj holds the new values and version. Onto(nil) writes without loading and locks only on the request's version field.
type Scalar ¶ added in v0.2.0
Scalar is what Min and Max read: a Number, or time.Time from a timestamp column.
type ScopeFunc ¶
ScopeFunc narrows every query of a store: it receives the query and returns it with more conditions. An error fails the operation closed.
type Store ¶
Store is the data access of model T on a database handle.
func New ¶
New returns the store of T on d. It panics on a declaration mistake, which surfaces at startup: a model without store tags, a store tag value other than query and update, an alias of an undeclared field, a hook for another model, a blank admin role or constraint.
func (*Store[T]) Avg ¶ added in v0.2.0
func (s *Store[T]) Avg(ctx context.Context, field string, opts ...where.Option) (float64, bool, error)
Avg returns the average of field, a numeric column, and false when there was nothing to average.
func (*Store[T]) BatchCreate ¶
BatchCreate inserts objs in one transaction, all or none.
func (*Store[T]) BatchUpsert ¶ added in v0.2.0
BatchUpsert is Upsert of objs in one transaction. Two objects with the same conflict values are an error, as the database would refuse them.
func (*Store[T]) Count ¶ added in v0.2.0
Count returns the number of rows the filters match; order and pagination options are ignored.
func (*Store[T]) CountDistinct ¶ added in v0.2.0
func (s *Store[T]) CountDistinct(ctx context.Context, field string, opts ...where.Option) (int64, error)
CountDistinct returns the number of distinct non-NULL values of field. Strings count by the column's collation, which may fold case.
func (*Store[T]) Create ¶
Create inserts obj and fills in its RID, version and timestamps. On an owned model the principal becomes the owner. A unique violation is a DuplicateError naming the field.
func (*Store[T]) Delete ¶
Delete removes the row loc names. A soft-delete model keeps the row with deleted_at set and a random delete_token, which frees its SoftUnique values; other models lose it. Deleting what is not there is not an error, except under where.Version: then a moved row is a VersionConflictError and a missing one a NotFoundError.
func (*Store[T]) Get ¶
Get returns the row loc names, or NotFoundError when there is none the caller may see. Options add filters, WithTrashed and Preload.
func (*Store[T]) GetForUpdate ¶ added in v0.2.0
GetForUpdate is Get that also locks the row until the transaction in ctx ends (SELECT ... FOR UPDATE), so no one else writes it in between; it fails outside a transaction. On SQLite the transaction already holds the only write connection, which is the stronger lock.
func (*Store[T]) GroupBy ¶ added in v0.2.0
func (s *Store[T]) GroupBy[K comparable](ctx context.Context, field string, aggs []Aggregate, opts ...where.Option) ([]Group[K], error)
GroupBy buckets the rows the filters match by field and computes aggs per bucket, ordered by key:
groups, err := orders.GroupBy[string](ctx, "status",
[]store.Aggregate{{Func: "count"}, {Func: "sum", Field: "amount"}})
K must match the field: string, int64, uint64, float64, bool or time.Time. A NULL key is an error (filter with where.NotNull). Order, pagination and Count are refused: they describe rows, not groups.
Example ¶
Aggregates are generic methods: the type argument says how to read the result, whatever the dialect returns.
package main
import (
"context"
"fmt"
"os"
"path/filepath"
"github.com/zynthara/chok"
"github.com/zynthara/chok/db"
"github.com/zynthara/chok/store"
)
// Article is an owned model: store tags declare what the API may filter
// on and update.
type Article struct {
db.OwnedSoftDeleteModel
Title string `json:"title" store:"query,update" gorm:"size:200;not null"`
Status string `json:"status" store:"query,update" gorm:"size:20;not null"`
}
func (Article) RIDPrefix() string { return "art" }
func openExample() (*db.DB, func()) {
dir, _ := os.MkdirTemp("", "chok-store-example")
d := db.MustOpen(db.Config{Driver: "sqlite", SQLite: db.SQLiteConfig{Path: filepath.Join(dir, "app.db")}},
db.Tables(db.Table(&Article{}, db.SoftUnique("uk_article_title_owner", "title", "owner_id"))))
return d, func() { _ = d.Close(context.Background()); _ = os.RemoveAll(dir) }
}
func main() {
d, done := openExample()
defer done()
articles := store.New[Article](d)
alice := chok.WithPrincipal(context.Background(), chok.Principal{UserID: "alice"})
for _, title := range []string{"a", "b", "c"} {
_ = articles.Create(alice, &Article{Title: title, Status: "draft"})
}
n, _ := articles.CountDistinct(alice, "title")
groups, _ := articles.GroupBy[string](alice, "status", []store.Aggregate{{Func: "count"}})
count, _ := groups[0].Values[0].Int64()
fmt.Println(n, groups[0].Key, count)
}
Output: 3 draft 3
func (*Store[T]) List ¶
List returns the rows the filters match, sorted by where.Order with the row ID breaking ties so pages never overlap. Without pagination options it returns every row, as one page. where.Page, or Limit and Offset, pick a page of the size they name, the policy's default page size when they name none; the policy's max bounds only the pagination of where.FromQuery, a client's request (ListFromQuery). Page is the page number — 1 without pagination options, 0 for Offset and Limit, which have none. HasMore says whether rows follow; Total is filled only with where.Count. With where.Cursor, List is ListWithCursor.
func (*Store[T]) ListByIDs ¶
ListByIDs is ListIn on the numeric primary key, which has no field name: the rows whose key is in ids, as PluckIDs or a foreign key column gives them, in key order.
func (*Store[T]) ListFromQuery ¶
ListFromQuery serves chok.List: page, size, order=field:asc|desc and cursor from the query string, and every query field as an equality filter (a repeated parameter matches any of its values). A filter on a boolean, numeric or time field is read as that type first: true or false (1 and 0 too), a number, an RFC 3339 time. An unknown parameter, or a value that does not parse, is a 400. The page is always one page: the policy's default page size when the query names none, and never more than its max. It always carries Total.
func (*Store[T]) ListIn ¶ added in v0.2.0
func (s *Store[T]) ListIn[V comparable](ctx context.Context, field string, values []V, opts ...where.Option) ([]T, error)
ListIn returns the rows whose field is one of values, querying 500 values at a time so any number fits; options may only filter.
func (*Store[T]) ListWithCursor ¶
ListWithCursor returns a keyset page: the rows after where.Cursor's cursor (none for the first page), ordered by the one where.Order given, or by ID. The page's Cursor continues it and is empty on the last page; clients pass it back untouched. Unlike offset pages, rows added or removed meanwhile never shift a cursor page. The page has the size where.Cursor names, the policy's default page size without one, and a cursor from where.FromQuery stays within the policy's max. The order field must be a NOT NULL scalar; Total is filled only with where.Count.
func (*Store[T]) Max ¶ added in v0.2.0
func (s *Store[T]) Max[N Scalar](ctx context.Context, field string, opts ...where.Option) (N, bool, error)
Max returns the largest value of field, as Min.
func (*Store[T]) Min ¶ added in v0.2.0
func (s *Store[T]) Min[N Scalar](ctx context.Context, field string, opts ...where.Option) (N, bool, error)
Min returns the smallest value of field, a numeric or time column, and false when there was none. Compare times with Equal: the zone differs by dialect.
func (*Store[T]) Pluck ¶ added in v0.2.0
func (s *Store[T]) Pluck[F any](ctx context.Context, field string, opts ...where.Option) ([]F, error)
Pluck returns field of the rows the filters match, in where.Order's order. Unlike List it takes every row unless where.Limit or where.Page says otherwise.
func (*Store[T]) PluckIDs ¶ added in v0.2.0
PluckIDs returns the numeric primary keys of the rows the filters match, for server-side work; they never leave the server.
func (*Store[T]) QueryFields ¶ added in v0.2.0
QueryFields lists the fields a URL filters and orders on: those ListFromQuery and chok.List accept, and the OpenAPI document lists.
func (*Store[T]) Restore ¶ added in v0.2.0
Restore brings back the soft-deleted row loc names. If a live row took one of its SoftUnique values meanwhile, Restore is a DuplicateError and the row stays deleted. Restoring a live row does nothing; a row the caller may not see is a NotFoundError.
func (*Store[T]) Sum ¶ added in v0.2.0
func (s *Store[T]) Sum[N Number](ctx context.Context, field string, opts ...where.Option) (N, bool, error)
Sum returns the sum of field over the rows the filters match, and false when there was nothing to sum (no rows, or only NULLs).
func (*Store[T]) Tx ¶
Tx runs fn in a transaction; the store's methods called with txCtx, and those of any store on the same handle, join it.
func (*Store[T]) Unsafe ¶ added in v0.2.0
Unsafe returns the *gorm.DB of the store's model with the owner scope and the custom scopes applied, joined to the transaction in ctx. A scope that fails leaves the error on the returned *gorm.DB, so whatever runs on it fails: the escape hatch is fenced like the rest. To get past the fences, use (*db.DB).Unsafe on the handle the store was built on.
func (*Store[T]) Update ¶
func (s *Store[T]) Update(ctx context.Context, loc Locator, ch *Changes, opts ...where.Option) error
Update writes ch to the row loc names and advances its version. The lock is where.Version, else the version Patch(...).Onto carried: when the row has moved past it, Update fails with VersionConflictError, and on success the object given to Onto holds the new values, version and updated_at, ready to return. A row the caller may not see is a NotFoundError.
func (*Store[T]) Upsert ¶
Upsert inserts obj or, when a row with the same conflict fields exists, overwrites that row's update fields and advances its version. conflict names the fields of a unique index, the RID when empty. On a soft-delete model a SoftUnique index counts only live rows; any other index, a Unique or the RID's, counts deleted rows too, and a deleted row holding obj's values is the DuplicateError Create gets, with nothing written. The check runs before the write: on PostgreSQL and MySQL a row deleted in between is still the one updated, and stays deleted; SQLite's single writer leaves no such gap. MySQL updates on any unique key whatever conflict says. obj is not reloaded: after an update it does not hold the stored row's RID or version. An owner-scoped or custom-scoped store refuses Upsert, whose update cannot be fenced: use Create, and Update on a DuplicateError.
type VersionConflictError ¶
VersionConflictError is an update or delete whose version no longer matches the row: someone changed it since it was read.
func (*VersionConflictError) APIError ¶ added in v0.2.0
func (e *VersionConflictError) APIError() *chok.Error
APIError is the 409, with metadata.version.
func (*VersionConflictError) Error ¶
func (e *VersionConflictError) Error() string
func (*VersionConflictError) Is ¶
func (e *VersionConflictError) Is(target error) bool
Is matches chok.ErrConflict.
Source Files
¶
Directories
¶
| Path | Synopsis |
|---|---|
|
internal
|
|
|
q
Package q holds what store and store/where share without exporting it: resolving a public field name to a column, safely.
|
Package q holds what store and store/where share without exporting it: resolving a public field name to a column, safely. |
|
Package where is the query vocabulary of store: filters, ordering, pagination and the query modifiers, each an Option that a store applies to its query.
|
Package where is the query vocabulary of store: filters, ordering, pagination and the query modifiers, each an Option that a store applies to its query. |