Allow multiple transaction extension version in `UncheckedExtrinsic` type. #7035

gui1117 · 2025-01-03T05:11:51Z

This PR enhance UncheckedExtrinsic type with a new optional generic: ExtensionOtherVersion.
This generic defaults to InvalidVersion meaning there is not other version than the regular version 0. This is the same behavior as before this PR.

New feature

To use this new feature, you can use the new types TxExtLineAtVers and MultiVersion to define a transaction extension with multiple version:

pub type TransactionExtensionV0 = ();
pub type TransactionExtensionV4 = ();
pub type TransactionExtensionV7 = ();

pub type OtherVersions = MultiVersion<
	TxExtLineAtVers<4, TransactionExtensionV4>;
	TxExtLineAtVers<7, TransactionExtensionV7>;
>;

pub type UncheckedExtrinsic = generic::UncheckedExtrinsic<
	AccountId,
	RuntimeCall,
	UintAuthorityId,
	TransactionExtensionV0, // The version 0, same as before
	OtherVersions, // The other versions.
>;

Breaking change

The types Preamble, CheckedExtrinsic and ExtrinsicFormat also have this new optional generic. Their type definition also have changed a bit, the General variant was 2 fields, the version and the extension, it is now only one field, the extension, and the version can be retrieve by calling extension.version()

The type inference for those types may fail because of this PR, to update the code, write some partial type: UncheckedExtrinsic<_, _, _, _>, Preamble<_, _, _>, ExtrinsicFormat<_, _> and CheckedExtrinsic<_, _, _>`.

Alternative implementation

This PR breaks the types a bit, I think it is very minimal and fine, but if this is annoying we can still keep the old types and write new types such as UncheckedExtrinsicV2, CheckedExtrinsicV2 etc..

…ension-multiple-version

franciscoaguirre

Looks good to me

prdoc/pr_7035.prdoc

substrate/frame/support/test/tests/tx_ext_multi_version.rs

franciscoaguirre · 2026-01-07T20:00:35Z

substrate/frame/support/test/tests/tx_ext_multi_version.rs

+}
+
+#[test]
+fn dispatch_of_invalid_extrinsic_fails() {


Could you add a comment explaining why this is invalid?

yes, I changed the inner comments to make it more clear. Also the previous inner comment was wrong.

Fixed: 0071ee2

substrate/primitives/runtime/src/generic/checked_extrinsic.rs

franciscoaguirre · 2026-01-07T20:08:39Z

substrate/primitives/runtime/src/traits/vers_tx_ext/mod.rs

+///
+/// This trait is part of [`VersTxExtLine`]. It is defined independently to allow implementation to
+/// rely only on it without bounding the whole trait [`VersTxExtLine`].
+pub trait VersTxExtLineWeight<Call: Dispatchable> {


What's the Line in the name?

It is short for pipeline. I couldn't find good names for this. The main issue is that we often conflate a singular transaction extension, and a pipeline of transaction extension. They are both represented in a single trait TransactionExtension where the associated const IDENTIFIER is only meaningful for the singular transaction extension, and the associated function fn metadata() -> Vec<TransactionExtensionMetadata> has a type signatured designed for the pipeline, and the automatic implementation is wrong for the pipeline and must be override.

So for those new types I tried to explicitly make a difference by adding this Line in the name, which means it is only meaningful for a pipeline.
The version we give here is only a version for the pipeline, not an individual transaction extension.

I added the reason for the name in the doc: 0071ee2

substrate/primitives/runtime/src/generic/unchecked_extrinsic.rs

carlosala

It looks good!

carlosala · 2026-01-08T10:24:54Z

substrate/primitives/metadata-ir/src/v15.rs

 			signature_ty: ir.signature_ty,
 			extra_ty: ir.extra_ty,
-			signed_extensions: ir.extensions.into_iter().map(Into::into).collect(),
+			signed_extensions: ir.extensions_v0.into_iter().map(Into::into).collect(),


Just passing by, checking this change was made, and we were not adding lots of wrong extensions!

yes and for metadata v16, if a user still want to use a signed transaction instead of a general transaction, the version for the the extensions is 0.

Co-authored-by: Francisco Aguirre <[email protected]>

bkchr · 2026-01-12T12:29:13Z

substrate/primitives/runtime/src/traits/mod.rs

+	/// The transaction extensions version 0 attached to this `Extrinsic`.
+	// We could remove this associated type and let user retrieve it from
+	// `TransactionExtensionsVersions`.
+	type TransactionExtensionsV0;


Yeah we should probably merge.

yes I now removed TransactionExtensionsV0: a49294c

bkchr · 2026-01-12T12:29:46Z

substrate/primitives/runtime/src/traits/mod.rs

+	type TransactionExtensionsV0;
+
+	/// All version of transaction extensions attached to this `Extrinsic`.
+	type TransactionExtensionsVersions;


Naming could be improved, because these are the transaction extensions and not just their versions?

I will use TransactionExtensionPipelines I think

bkchr · 2026-01-12T13:31:03Z

substrate/primitives/runtime/src/generic/unchecked_extrinsic.rs

+				// NOTE: We avoid adding `ExtensionOtherVersions` in the type parameter type info
+				// because they break tools like subxt that hard-coded the number of type
+				// parameters. Also in general type parameters are not useful, information related
+				// to transaction extension are available in the metadata like for v16 the fields
+				// `transaction_extensions_by_version` and `transaction_extensions`.


What is subxt using internally to build/verify the extrinsic? The dedicated extrinsic metadata or just this type somehow extracted from the metadata?

When reading the type info, they replace some types in primitives to their own primitives written here: https://github.com/paritytech/subxt/blob/a2080bf1f7414e3e55429984b3d53f3828fe7897/core/src/utils/unchecked_extrinsic.rs#L20-L25

And the default substitutes are written here: https://github.com/paritytech/subxt/blob/a2080bf1f7414e3e55429984b3d53f3828fe7897/codegen/src/lib.rs#L428-L432

bkchr · 2026-01-12T13:32:26Z

substrate/primitives/metadata-ir/src/types.rs

+	/// The transaction extensions in the order they appear in the extrinsic for the version 0.
+	pub extensions_v0: Vec<TransactionExtensionMetadataIR<T>>,


IMO we should remove this and have the invariant that 0 always exists in extensions_by_version.

yes I now removed TransactionExtensionsV0: a9942a8

bkchr · 2026-01-13T09:18:54Z

substrate/primitives/metadata-ir/src/types.rs

+	/// The transaction extensions in the order they appear in the extrinsic for the version 0.
+	pub extensions_v0: Vec<TransactionExtensionMetadataIR<T>>,


Yeah I would remove this and always assume that extensions_by_version contains 0.

yes I now removed TransactionExtensionsV0: a9942a8

bkchr · 2026-01-13T09:59:47Z

substrate/primitives/runtime/src/traits/vers_tx_ext/invalid.rs

+		// The type cannot be instantiated so this method is never called.
+		Err(TransactionValidityError::Invalid(InvalidTransaction::Custom(0)))


Then just use unimplemented!()

ok fixed here e071249

bkchr · 2026-01-13T10:16:24Z

substrate/primitives/runtime/src/traits/vers_tx_ext/mod.rs

+
+/// A type implements [`DecodeWithVersion`] where inner decoding is implementing
+/// [`codec::DecodeWithMemTracking`].
+pub trait DecodeWithVersionWithMemTracking: DecodeWithVersion {}


Yeah why do we need this?

bkchr · 2026-01-13T10:17:05Z

substrate/primitives/runtime/src/traits/vers_tx_ext/mod.rs

+/// This defines multiple version of a transaction extensions pipeline.
+///
+/// (type name is short for versioned (Vers) transaction (Tx) Extension (Ext) pipeline (Line)).
+pub trait VersTxExtLine<Call: Dispatchable>:


Why do we need 5 different traits? Can this not just be one trait?

For VersTxExtLineWeight, we could merge it in, I avoided because it required to change some bounds here and there that weren't bound before. But also it looks ok to force them as I don't think anybody use the genericity offered here. https://github.com/paritytech/polkadot-sdk/compare/gui-transaction-extension-multiple-version...gui-transaction-extension-multiple-version-remove-tx-weight?expand=1

For VersTxExtLineVersion, I don't think we can merge it in. It it required to derive Encode for many types such as Preamble but VersTxExtLine is generic over the call, and Preamble doesn't have a call.
Then the question is why VersTxExtLine is generic over the call? IIRC because TransactionExtension is generic over the call. Why is TransactionExtension generic over the call? actually I think for no reason. Usually the transaction extensions are defined generically over the runtime config, which defines the call. So TransactionExtension could be not generic and instead have an associated type name Call but this is a bit late. Or Maybe I miss something.

For DecodeWithVersion, same issue as VersTxExtLineVersion but for implementing Decode on Preamble.

(Note that maybe both issue could be fixed with having Preamble use a phantom data for the call. not sure).

For DecodeWithVersionWithMemTracking, for consistency and hopefully less errors, but also I agree we can merge into DecodeWithVersion.

bkchr · 2026-01-13T10:24:14Z

substrate/primitives/runtime/src/traits/vers_tx_ext/multi.rs

+		/// ```
+		#[allow(private_interfaces)]
+		#[derive(PartialEq, Eq, Clone, Debug, TypeInfo)]
+		pub enum MultiVersion<


This entire type can just be replaced by a tuple of TxExtLineAtVers?

We can replace it with a tuple of something I guess but not TxExtLineAtVers. We need to implement Encode and Decode, so the resulting type is not a tuple of all possible pipelines but it is the variant for the version.

If we want to define this aggregation of versioned pipeline with a tuple I guess we need a new trait that decodes into a trait object?

trait PipelineDefinition { fn decode() -> Result<Box<impl VersTxExtLine>, ()>; fn metadata() -> Metadata; }

Then we can define all the supported version with a tuple of PipelineDefinition.
If we want this we also need to move the type info into this PipelineDefinition because it would make the VersTxExtLine incompatible with trait object usage.

Maybe MultiVersions is ok.

or we can make a tuple that defines an associated type and the associated type is just MultiVersion of each element of the tuple's associated type.

bkchr · 2026-01-13T10:29:10Z

substrate/primitives/runtime/src/traits/mod.rs

+	// We could remove this associated type and let user retrieve it from
+	// `TransactionExtensionsVersions`.
+	type TransactionExtensionsV0;
+
+	/// All version of transaction extensions attached to this `Extrinsic`.
+	type TransactionExtensionsVersions;


Probably also not worth the split here?

yes I now removed TransactionExtensionsV0: a49294c

gui1117 · 2026-01-25T10:33:50Z

I currently renamed into:

VersTxExtLine → Pipeline : The transaction extension pipeline global trait
VersTxExtLineWeight → PipelineWeight : The weight trait for the transaction extension pipeline (probably we should merge this into Pipeline)
VersTxExtLineVersion → PipelineVersion : The version trait for the transaction extension pipeline
VersTxExtLineMetadataBuilder → PipelineMetadataBuilder : Just a helper type to build the metadata, not so important IMO
TxExtLineAtVers → PipelineAtVers : The type to easily declare a pipeline for a specific version. Maybe AtVers is a better name?
TransactionExtensionsVersions → TransactionExtensionPipelines : the associated type in the metadata for the definition of all transaction extension pipelines

paritytech-workflow-stopper · 2026-01-27T03:01:20Z

All GitHub workflows were cancelled due to failure one of the required jobs.
Failed workflow url: https://github.com/paritytech/polkadot-sdk/actions/runs/21381994474
Failed job name: test-linux-stable-no-try-runtime

gui1117 added 27 commits December 12, 2024 20:05

WIP

b5c95b1

WIP

582cbb9

WIP

32159f9

WIP

2350e93

WIP

143b50a

cleanup

17a8dc0

WIP

a38fac3

WIP

3f657ac

Structure done: todo: renames and format

06da89b

Merge remote-tracking branch 'origin/master' into gui-transaction-ext…

b03b4eb

…ension-multiple-version

renames

cfddfa1

reorganize

3aacd46

fixes

e060cac

description draft

2c1e2fc

Merge remote-tracking branch 'origin/master' into gui-transaction-ext…

d39156e

…ension-multiple-version

refactor

0cc8f0c

doc

049ec5d

tests + fmt

d9eb5b8

multi version support 26 versions

8e32689

simplify bounds

c6aa890

refactor test

bfadf05

remove unneeded bounds

dc7944f

more tests

ad9e9ba

make invalid version safer

1812860

Merge remote-tracking branch 'origin/master' into gui-transaction-ext…

9967ab0

…ension-multiple-version

fmt

082ab90

prdoc

a4a87f5

gui1117 marked this pull request as ready for review January 8, 2025 09:44

gui1117 requested a review from a team as a code owner January 8, 2025 09:44

gui1117 added the T1-FRAME This PR/Issue is related to core FRAME, the framework. label Jan 8, 2025

gui1117 added 2 commits January 1, 2026 13:09

better type info

4b6ef76

better rename + impl revive

31ea43d

paritytech deleted a comment from paritytech-workflow-stopper bot Jan 1, 2026

gui1117 added 2 commits January 1, 2026 13:58

prdoc

b2621eb

fix revive

8284dc8

paritytech deleted a comment from paritytech-workflow-stopper bot Jan 1, 2026

franciscoaguirre approved these changes Jan 7, 2026

View reviewed changes

carlosala reviewed Jan 8, 2026

View reviewed changes

carlosala mentioned this pull request Jan 8, 2026

fix(signer): only check relevant signed extensions polkadot-api/polkadot-api#1274

Merged

gui1117 and others added 5 commits January 9, 2026 12:37

Apply suggestions from code review

acf6904

Co-authored-by: Francisco Aguirre <[email protected]>

Update substrate/frame/support/test/tests/tx_ext_multi_version.rs

4b08c49

Co-authored-by: Francisco Aguirre <[email protected]>

Update substrate/primitives/runtime/src/generic/unchecked_extrinsic.rs

316e623

Co-authored-by: Francisco Aguirre <[email protected]>

Apply suggestion from @franciscoaguirre

64cfd4c

Co-authored-by: Francisco Aguirre <[email protected]>

better doc and comments

0071ee2

bkchr reviewed Jan 13, 2026

View reviewed changes

gui1117 added 3 commits January 22, 2026 15:17

use unreachable

e071249

remove metadata ir extension v0 type

a9942a8

remove ExtensionV0 type in metadata

a49294c

gui1117 and others added 3 commits January 25, 2026 19:35

rename

5bb9d50

fix unused

fcafaac

Merge branch 'master' into gui-transaction-extension-multiple-version

7a6eaf0

paritytech deleted a comment from paritytech-workflow-stopper bot Jan 25, 2026

fmt

1879c54

paritytech deleted a comment from paritytech-workflow-stopper bot Jan 27, 2026

		/// The transaction extensions in the order they appear in the extrinsic for the version 0.
		pub extensions_v0: Vec<TransactionExtensionMetadataIR<T>>,

		// The type cannot be instantiated so this method is never called.
		Err(TransactionValidityError::Invalid(InvalidTransaction::Custom(0)))

Allow multiple transaction extension version in UncheckedExtrinsic type. #7035

Are you sure you want to change the base?

Allow multiple transaction extension version in UncheckedExtrinsic type. #7035

Uh oh!

Conversation

gui1117 commented Jan 3, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

New feature

Breaking change

Alternative implementation

Uh oh!

franciscoaguirre left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

gui1117 Jan 9, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

carlosala left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

gui1117 Jan 22, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

gui1117 Jan 22, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

gui1117 Jan 22, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Allow multiple transaction extension version in `UncheckedExtrinsic` type. #7035

Allow multiple transaction extension version in `UncheckedExtrinsic` type. #7035

gui1117 commented Jan 3, 2025 •

edited

Loading

gui1117 Jan 9, 2026 •

edited

Loading

gui1117 Jan 22, 2026 •

edited

Loading

gui1117 Jan 22, 2026 •

edited

Loading

gui1117 Jan 22, 2026 •

edited

Loading

gui1117 Jan 22, 2026 •

edited

Loading