inhji/hajur

Watch

Fork 0

chore(deps): update dependency flop to ~> 0.29.0 #235

Open

renovate wants to merge 1 commit from renovate/flop-0.x into main

renovate commented

2026-09-06 12:00:42 +02:00

Collaborator

This PR contains the following updates:

Package	Type	Update	Change
flop (source)	prod	minor	`~> 0.26.4` → `~> 0.29.0`

Release Notes

woylie/flop (flop)

`v0.29.0`

Compare Source

Added

Add Flop.allowed_fields/2, which returns the fields that may be filtered or
sorted for the given options.
Add Flop.Schema.flop_schema!/1, which returns the configuration of a schema
module.
Add Flop.Cursor.get_cursor_from_node/3 and
Flop.Cursor.get_cursor_from_edge/3, which take the query options and resolve
a field against the schema given as :for.
Accept an arity-3 cursor_value_func, which also receives the options the
query ran with.

Changed

Turn Flop.Schema from a protocol into a behaviour. Schemas are configured
with use Flop.Schema and the @flop_options attribute now.
Take the schema module instead of a struct in Flop.Schema.field_info/2,
Flop.Schema.get_field/3 and Flop.Schema.primary_key/1.
Default cursor_value_func to Flop.Cursor.get_cursor_from_node/3.

Removed

Remove default_limit/1, default_order/1, default_pagination_type/1,
filterable/1, max_filters/1, max_limit/1, pagination_types/1,
sortable/1 and tiebreaker/1 from Flop.Schema.

Fixed

Do not make a module named in an ecto_type a compile-time dependency of the
schema.
Read a cursor value through the schema given as :for, so that a field's
:path is applied when the query returns plain maps rather than structs.

How to upgrade

Change your code to use Flop.Schema and set the options as @flop_options
instead of using @derive to derive the protocol.

 defmodule MyApp.Pet do
   use Ecto.Schema
+  use Flop.Schema

-  @&#8203;derive {
-    Flop.Schema,
-    filterable: [:name, :species],
-    sortable: [:name, :age]
-  }
+  @&#8203;flop_options [
+    filterable: [:name, :species],
+    sortable: [:name, :age]
+  ]

   schema "pets" do
     field :name, :string
     field :age, :integer
     field :species, :string
   end
 end

The attribute is read when the module finishes compiling, so it can go anywhere
in the module body. If you use an alias in the options, the alias has to be
declared before the module attribute.

The functions that returned a single schema option are gone. Read the effective
value through Flop.get_option/3, which also applies the backend module and the
application environment, or Flop.allowed_fields/2 for filterable and
sortable, which applies the narrowing a query function may have set:

-Flop.Schema.sortable(%MyApp.Pet{})
+Flop.allowed_fields(:sortable, for: MyApp.Pet)

-Flop.Schema.max_limit(%MyApp.Pet{})
+Flop.get_option(:max_limit, for: MyApp.Pet)

The remaining functions take the schema module instead of a struct:

-Flop.Schema.field_info(%MyApp.Pet{}, :name)
+Flop.Schema.field_info(MyApp.Pet, :name)

-Flop.Schema.get_field(pet, :owner_name)
+Flop.Schema.get_field(MyApp.Pet, pet, :owner_name)

An arity-2 cursor_value_func is still supported. You can use a 3-arity
function if you need the schema or other options passed to Flop, for example if
you need to apply the path option of join or custom fields.

-cursor_value_func: fn item, fields -> Map.take(item, fields) end
+cursor_value_func: fn item, fields, opts ->
+  Map.new(fields, &{&1, Flop.Schema.get_field(opts[:for], item, &1)})
+end

`v0.28.0`

Compare Source

Added

Support SQLite.
Support MySQL.
Support ordering, filtering and cursor pagination by custom fields via a new
field_dynamic option.
Support the path option on custom fields.
Add the tiebreaker option, which appends order fields to make the order
stable.
Add Flop.ordering/2 and Flop.cursor_fields/2, which return the order that
is applied to a query.
Add the diagnostics option, which enables checks about how Flop is used. It
can only be set in the application environment.
Allow setting max_filters when deriving Flop.Schema.

Changed

Raise when a cursor is applied with a plain :asc or :desc order direction
and no repo is configured. This only affects queries built with
Flop.query/3.
Reject ecto_type: nil on join and custom fields.
Append the primary key to the order Flop applies by default. The fields to use
can be overridden with the tiebreaker option. Set tiebreaker: false to
disable.
Accept cursor pagination parameters without an order field unless the
tiebreaker is disabled.
Add tiebreaker/1, primary_key/1 and max_filters/1 to the Flop.Schema
protocol.
Only log a warning about a query that already has an ORDER BY clause when
the diagnostics option is enabled in the application environment.
Allow to pass filterable as an option to query functions to narrow the
schema's filterable fields.
Apply the sortable option passed to a query function by narrowing the
schema's list instead of replacing it.
Raise an ArgumentError when a Flop struct with a non-matching pagination
type is passed to Flop.to_previous_page/1, Flop.to_next_page/2,
Flop.to_previous_offset/1 or Flop.to_next_offset/2, and apply them to the
first page when it has no pagination parameters.

Fixed

Build the :=~, :ilike, :not_ilike, :ilike_and, :ilike_or,
:starts_with and :ends_with filters with LIKE on databases that have no
ILIKE, such as SQLite and MySQL, instead of raising an Ecto.QueryError.
Support :asc_nulls_first, :asc_nulls_last, :desc_nulls_first and
:desc_nulls_last directions for MySQL.
Build the :contains and :not_contains filters, and :empty/:not_empty
on array fields, with JSON_CONTAINS and JSON_LENGTH on MySQL, instead of
raising.
Remove the :=~ operator from the operators allowed for boolean fields.
Remove the pattern operators (:like, :starts_with, etc.) from the
operators allowed for binary and binary_id fields.
Remove the :contains and :not_contains operators from the operators
allowed for binary fields.
Fix the partial UUID filter recipe, which used ilike and raised on MySQL and
SQLite, and add the variant MySQL needs. Note in the conditional joins recipe
that SQLite has no lateral joins.
Support cursor pagination on nullable columns.
Return a validation error instead of raising a FunctionClauseError when a
join field is configured with the tuple shorthand that was removed earlier.
Ignore sortable, filterable, for, count, count_query and
extra_opts set in the application environment. The application environment
now accepts the same options as a backend module.
Raise instead of returning nil when Flop.Schema.get_field/2 reads a value
through an association that is not loaded.

How to upgrade

The warning about a query that already has an ORDER BY clause is now only
logged when diagnostics are enabled. To keep seeing it, add this to
config/dev.exs and config/test.exs:

config :flop, diagnostics: true

sortable and filterable passed to a query function now narrow the schema's
lists instead of replacing them. A call site can no longer add a field the
schema does not list. Add any missing field to the schema options:

 @&#8203;derive {Flop.Schema,
   filterable: [:name],
-  sortable: [:name]}
+  sortable: [:name, :internal_rank]}

 Flop.validate_and_run(Pet, params, for: Pet, sortable: [:internal_rank])

Flop appends the primary key to the order now by default, which means that the
field list passed to cursor_value_func includes it now, and its value needs to
be in the returned map. Ensure your custom cursor_value_func can handle the
additional field.

-cursor_value_func: fn item, [:name] -> %{name: item.name} end
+cursor_value_func: fn item, fields -> Map.take(item, fields) end

You can set tiebreaker: false to disable this feature and revert to the
previous behaviour.

`v0.27.2`

Compare Source

Added

Add a guide for case-insensitive sorting.

Fixed

Apply the :empty and :not_empty filters on SQLite correctly.
Add an ESCAPE clause to LIKE patterns, so that filter values containing
%, _ or \ are matched literally on databases that define no default
escape character, such as SQLite.
Fix :empty and :not_empty filters with the value false on compound
fields. :empty with false required every field to be set, instead of any
field, so it did not return the same rows as :not_empty with true.

`v0.27.1`

Compare Source

Added

Add several new guides to the documentation.

Fixed

Reject filter and cursor values that contain NUL bytes or invalid UTF-8.
Resolve the replace_invalid_params option through the documented look-up
order. It was only applied when passed to a query function, and can now be set
on a backend module and in the application environment as well.
Resolve the cursor_value_func option through the documented look-up order
and accept it in use Flop.
Accept the max_cursor_size option in use Flop.
Accept false for the max_limit option in use Flop and in
@derive Flop.Schema.
Ignore a default_order set in the application environment.

`v0.27.0`

Compare Source

Changed

Require the ecto_type option for join fields and custom fields.
Raise a Flop.UnknownFieldError instead of a FunctionClauseError when
Flop.Schema.field_info/2, Flop.Schema.get_field/2 or
Flop.Filter.allowed_operators/2 is called with a field that is not
configured in the schema.
Reject compound and alias fields as order fields for cursor pagination during
validation. They can still be used with offset and page based pagination.
Raise an ArgumentError instead of ignoring the filter when an unsupported
operator is applied to a compound field. This only affects unvalidated
filters.

Fixed

Return an empty list from Flop.aliases/2 if the order_by parameter is
nil, instead of raising a Protocol.UndefinedError.
Build the meta struct in Flop.meta/3 if neither a page size nor a limit is
set, instead of raising an ArithmeticError.
Fall back to the first page or to offset 0 in Flop.set_page/2 and
Flop.set_offset/2 if the given string is not a number, instead of raising
an ArgumentError.
Merge adapter options passed at the call site with the options set on the
backend module, instead of discarding them.

Security

Limit the number of filters that can be applied in one request to 20 by
default. Configurable with the max_filters option.
Remove the debug log on validation errors. It logged the changeset, which
contains the user-supplied filter values. The errors are returned to the
caller in the Flop.Meta struct, so nothing is lost.

How to upgrade

Add the ecto_type options to all Flop.Schema derivations that don't have it
already:

  @&#8203;derive {
    Flop.Schema,
    filterable: [],
    sortable: [],
    adapter_opts: [
      join_fields: [
        owner_age: [
          binding: :owner,
          field: :age,
+         ecto_type: :integer
        ]
      ],
      custom_fields: [
        partial_id: [
          filter: {__MODULE__, :partial_id_filter, []},
+         ecto_type: :string
        ]
      ]
    ]
  }

Filter lists longer than 20 are now rejected with a validation error on
filters. If your query handles longer filter lists, pass the max_filters
option:

Flop.validate_and_run(Pet, params, for: Pet, max_filters: 50)

The option can also be set via application environment or backend module.

Configuration

📅 Schedule: (UTC)

Branch creation
- At any time (no schedule defined)
Automerge
- At any time (no schedule defined)

🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.

♻ Rebasing: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox.

🔕 Ignore: Close this PR and you won't be reminded about this update again.

If you want to rebase/retry this PR, check this box

This PR has been generated by Mend Renovate.

This PR contains the following updates: | Package | Type | Update | Change | |---|---|---|---| | [flop](https://hex.pm/packages/flop) ([source](https://github.com/woylie/flop)) | prod | minor | `~> 0.26.4` → `~> 0.29.0` | --- ### Release Notes <details> <summary>woylie/flop (flop)</summary> ### [`v0.29.0`](https://github.com/woylie/flop/blob/HEAD/CHANGELOG.md#0290---2026-09-03) [Compare Source](https://github.com/woylie/flop/compare/0.28.0...0.29.0) ##### Added - Add `Flop.allowed_fields/2`, which returns the fields that may be filtered or sorted for the given options. - Add `Flop.Schema.flop_schema!/1`, which returns the configuration of a schema module. - Add `Flop.Cursor.get_cursor_from_node/3` and `Flop.Cursor.get_cursor_from_edge/3`, which take the query options and resolve a field against the schema given as `:for`. - Accept an arity-3 `cursor_value_func`, which also receives the options the query ran with. ##### Changed - Turn `Flop.Schema` from a protocol into a behaviour. Schemas are configured with `use Flop.Schema` and the `@flop_options` attribute now. - Take the schema module instead of a struct in `Flop.Schema.field_info/2`, `Flop.Schema.get_field/3` and `Flop.Schema.primary_key/1`. - Default `cursor_value_func` to `Flop.Cursor.get_cursor_from_node/3`. ##### Removed - Remove `default_limit/1`, `default_order/1`, `default_pagination_type/1`, `filterable/1`, `max_filters/1`, `max_limit/1`, `pagination_types/1`, `sortable/1` and `tiebreaker/1` from `Flop.Schema`. ##### Fixed - Do not make a module named in an `ecto_type` a compile-time dependency of the schema. - Read a cursor value through the schema given as `:for`, so that a field's `:path` is applied when the query returns plain maps rather than structs. ##### How to upgrade Change your code to `use Flop.Schema` and set the options as `@flop_options` instead of using `@derive` to derive the protocol. ```diff defmodule MyApp.Pet do use Ecto.Schema + use Flop.Schema - @derive { - Flop.Schema, - filterable: [:name, :species], - sortable: [:name, :age] - } + @flop_options [ + filterable: [:name, :species], + sortable: [:name, :age] + ] schema "pets" do field :name, :string field :age, :integer field :species, :string end end ``` The attribute is read when the module finishes compiling, so it can go anywhere in the module body. If you use an alias in the options, the alias has to be declared before the module attribute. The functions that returned a single schema option are gone. Read the effective value through `Flop.get_option/3`, which also applies the backend module and the application environment, or `Flop.allowed_fields/2` for `filterable` and `sortable`, which applies the narrowing a query function may have set: ```diff -Flop.Schema.sortable(%MyApp.Pet{}) +Flop.allowed_fields(:sortable, for: MyApp.Pet) -Flop.Schema.max_limit(%MyApp.Pet{}) +Flop.get_option(:max_limit, for: MyApp.Pet) ``` The remaining functions take the schema module instead of a struct: ```diff -Flop.Schema.field_info(%MyApp.Pet{}, :name) +Flop.Schema.field_info(MyApp.Pet, :name) -Flop.Schema.get_field(pet, :owner_name) +Flop.Schema.get_field(MyApp.Pet, pet, :owner_name) ``` An arity-2 `cursor_value_func` is still supported. You can use a 3-arity function if you need the schema or other options passed to Flop, for example if you need to apply the `path` option of join or custom fields. ```diff -cursor_value_func: fn item, fields -> Map.take(item, fields) end +cursor_value_func: fn item, fields, opts -> + Map.new(fields, &{&1, Flop.Schema.get_field(opts[:for], item, &1)}) +end ``` ### [`v0.28.0`](https://github.com/woylie/flop/blob/HEAD/CHANGELOG.md#0280---2026-08-24) [Compare Source](https://github.com/woylie/flop/compare/0.27.2...0.28.0) ##### Added - Support SQLite. - Support MySQL. - Support ordering, filtering and cursor pagination by custom fields via a new `field_dynamic` option. - Support the `path` option on custom fields. - Add the `tiebreaker` option, which appends order fields to make the order stable. - Add `Flop.ordering/2` and `Flop.cursor_fields/2`, which return the order that is applied to a query. - Add the `diagnostics` option, which enables checks about how Flop is used. It can only be set in the application environment. - Allow setting `max_filters` when deriving `Flop.Schema`. ##### Changed - Raise when a cursor is applied with a plain `:asc` or `:desc` order direction and no repo is configured. This only affects queries built with `Flop.query/3`. - Reject `ecto_type: nil` on join and custom fields. - Append the primary key to the order Flop applies by default. The fields to use can be overridden with the `tiebreaker` option. Set `tiebreaker: false` to disable. - Accept cursor pagination parameters without an order field unless the tiebreaker is disabled. - Add `tiebreaker/1`, `primary_key/1` and `max_filters/1` to the `Flop.Schema` protocol. - Only log a warning about a query that already has an `ORDER BY` clause when the `diagnostics` option is enabled in the application environment. - Allow to pass `filterable` as an option to query functions to narrow the schema's filterable fields. - Apply the `sortable` option passed to a query function by narrowing the schema's list instead of replacing it. - Raise an `ArgumentError` when a `Flop` struct with a non-matching pagination type is passed to `Flop.to_previous_page/1`, `Flop.to_next_page/2`, `Flop.to_previous_offset/1` or `Flop.to_next_offset/2`, and apply them to the first page when it has no pagination parameters. ##### Fixed - Build the `:=~`, `:ilike`, `:not_ilike`, `:ilike_and`, `:ilike_or`, `:starts_with` and `:ends_with` filters with `LIKE` on databases that have no `ILIKE`, such as SQLite and MySQL, instead of raising an `Ecto.QueryError`. - Support `:asc_nulls_first`, `:asc_nulls_last`, `:desc_nulls_first` and `:desc_nulls_last` directions for MySQL. - Build the `:contains` and `:not_contains` filters, and `:empty`/`:not_empty` on array fields, with `JSON_CONTAINS` and `JSON_LENGTH` on MySQL, instead of raising. - Remove the `:=~` operator from the operators allowed for boolean fields. - Remove the pattern operators (`:like`, `:starts_with`, etc.) from the operators allowed for `binary` and `binary_id` fields. - Remove the `:contains` and `:not_contains` operators from the operators allowed for `binary` fields. - Fix the partial UUID filter recipe, which used `ilike` and raised on MySQL and SQLite, and add the variant MySQL needs. Note in the conditional joins recipe that SQLite has no lateral joins. - Support cursor pagination on nullable columns. - Return a validation error instead of raising a `FunctionClauseError` when a join field is configured with the tuple shorthand that was removed earlier. - Ignore `sortable`, `filterable`, `for`, `count`, `count_query` and `extra_opts` set in the application environment. The application environment now accepts the same options as a backend module. - Raise instead of returning `nil` when `Flop.Schema.get_field/2` reads a value through an association that is not loaded. ##### How to upgrade The warning about a query that already has an `ORDER BY` clause is now only logged when diagnostics are enabled. To keep seeing it, add this to `config/dev.exs` and `config/test.exs`: ```elixir config :flop, diagnostics: true ``` `sortable` and `filterable` passed to a query function now narrow the schema's lists instead of replacing them. A call site can no longer add a field the schema does not list. Add any missing field to the schema options: ```diff @derive {Flop.Schema, filterable: [:name], - sortable: [:name]} + sortable: [:name, :internal_rank]} Flop.validate_and_run(Pet, params, for: Pet, sortable: [:internal_rank]) ``` Flop appends the primary key to the order now by default, which means that the field list passed to `cursor_value_func` includes it now, and its value needs to be in the returned map. Ensure your custom `cursor_value_func` can handle the additional field. ```diff -cursor_value_func: fn item, [:name] -> %{name: item.name} end +cursor_value_func: fn item, fields -> Map.take(item, fields) end ``` You can set `tiebreaker: false` to disable this feature and revert to the previous behaviour. ### [`v0.27.2`](https://github.com/woylie/flop/blob/HEAD/CHANGELOG.md#0272---2026-08-12) [Compare Source](https://github.com/woylie/flop/compare/0.27.1...0.27.2) ##### Added - Add a guide for case-insensitive sorting. ##### Fixed - Apply the `:empty` and `:not_empty` filters on SQLite correctly. - Add an `ESCAPE` clause to `LIKE` patterns, so that filter values containing `%`, `_` or `\` are matched literally on databases that define no default escape character, such as SQLite. - Fix `:empty` and `:not_empty` filters with the value `false` on compound fields. `:empty` with `false` required every field to be set, instead of any field, so it did not return the same rows as `:not_empty` with `true`. ### [`v0.27.1`](https://github.com/woylie/flop/blob/HEAD/CHANGELOG.md#0271---2026-08-11) [Compare Source](https://github.com/woylie/flop/compare/0.27.0...0.27.1) ##### Added - Add several new guides to the documentation. ##### Fixed - Reject filter and cursor values that contain NUL bytes or invalid UTF-8. - Resolve the `replace_invalid_params` option through the documented look-up order. It was only applied when passed to a query function, and can now be set on a backend module and in the application environment as well. - Resolve the `cursor_value_func` option through the documented look-up order and accept it in `use Flop`. - Accept the `max_cursor_size` option in `use Flop`. - Accept `false` for the `max_limit` option in `use Flop` and in `@derive Flop.Schema`. - Ignore a `default_order` set in the application environment. ### [`v0.27.0`](https://github.com/woylie/flop/blob/HEAD/CHANGELOG.md#0270---2026-08-10) [Compare Source](https://github.com/woylie/flop/compare/0.26.6...0.27.0) ##### Changed - Require the `ecto_type` option for join fields and custom fields. - Raise a `Flop.UnknownFieldError` instead of a `FunctionClauseError` when `Flop.Schema.field_info/2`, `Flop.Schema.get_field/2` or `Flop.Filter.allowed_operators/2` is called with a field that is not configured in the schema. - Reject compound and alias fields as order fields for cursor pagination during validation. They can still be used with offset and page based pagination. - Raise an `ArgumentError` instead of ignoring the filter when an unsupported operator is applied to a compound field. This only affects unvalidated filters. ##### Fixed - Return an empty list from `Flop.aliases/2` if the `order_by` parameter is `nil`, instead of raising a `Protocol.UndefinedError`. - Build the meta struct in `Flop.meta/3` if neither a page size nor a limit is set, instead of raising an `ArithmeticError`. - Fall back to the first page or to offset `0` in `Flop.set_page/2` and `Flop.set_offset/2` if the given string is not a number, instead of raising an `ArgumentError`. - Merge adapter options passed at the call site with the options set on the backend module, instead of discarding them. ##### Security - Limit the number of filters that can be applied in one request to 20 by default. Configurable with the `max_filters` option. - Remove the debug log on validation errors. It logged the changeset, which contains the user-supplied filter values. The errors are returned to the caller in the `Flop.Meta` struct, so nothing is lost. ##### How to upgrade Add the `ecto_type` options to all `Flop.Schema` derivations that don't have it already: ```diff @derive { Flop.Schema, filterable: [], sortable: [], adapter_opts: [ join_fields: [ owner_age: [ binding: :owner, field: :age, + ecto_type: :integer ] ], custom_fields: [ partial_id: [ filter: {__MODULE__, :partial_id_filter, []}, + ecto_type: :string ] ] ] } ``` Filter lists longer than 20 are now rejected with a validation error on `filters`. If your query handles longer filter lists, pass the `max_filters` option: ```elixir Flop.validate_and_run(Pet, params, for: Pet, max_filters: 50) ``` The option can also be set via application environment or backend module. </details> --- ### Configuration 📅 **Schedule**: (UTC) - Branch creation - At any time (no schedule defined) - Automerge - At any time (no schedule defined) 🚦 **Automerge**: Disabled by config. Please merge this manually once you are satisfied. ♻ **Rebasing**: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox. 🔕 **Ignore**: Close this PR and you won't be reminded about this update again. --- - [ ] If you want to rebase/retry this PR, check this box --- This PR has been generated by [Mend Renovate](https://github.com/renovatebot/renovate).

renovate added 1 commit

2026-09-06 12:00:42 +02:00

chore(deps): update dependency flop to ~> 0.29.0

renovate/artifacts Artifact file update failure

3c77b4fc62

renovate commented

2026-09-06 12:00:43 +02:00

Author

Collaborator

⚠️ Artifact update problem

Renovate failed to update an artifact related to this branch. You probably do not want to merge this PR as-is.

♻ Renovate will retry this branch, including artifacts, only when one of the following happens:

any of the package files in this branch needs updating, or
the branch becomes conflicted, or
you click the rebase/retry checkbox if found above, or
you rename this PR's title to start with "rebase!" to trigger it manually

The artifact failure details are included below:

File name: mix.lock

Command failed: mix deps.update flop
** (Mix) Hex dependency resolution failed

### ⚠️ Artifact update problem Renovate failed to update an artifact related to this branch. You probably do not want to merge this PR as-is. ♻ Renovate will retry this branch, including artifacts, only when one of the following happens: - any of the package files in this branch needs updating, or - the branch becomes conflicted, or - you click the rebase/retry checkbox if found above, or - you rename this PR's title to start with "rebase!" to trigger it manually The artifact failure details are included below: ##### File name: mix.lock ``` Command failed: mix deps.update flop ** (Mix) Hex dependency resolution failed ```

renovate changed title from ~~chore(deps): update dependency flop to ~> 0.29.0~~ to Update dependency flop to ~> 0.29.0

2026-09-07 12:00:33 +02:00