Remove ServiceMsgs from ADR-031 #9139

amaury1093 · 2021-04-19T10:19:37Z

Description

completely remove ServiceMsg TypeURLs, without supporting backwards compatbility
emit events with leading /, e.g. message.action='/cosmos.bank.v1beta1.MsgSend'

closes: #9063
closes: #8864
closes: #9172

Before we can merge this PR, please make sure that all the following items have been
checked off. If any of the checklist items are not applicable, please leave them but
write a little note why.

Targeted PR against correct branch (see CONTRIBUTING.md)
Linked to Github issue with discussion and accepted design OR link to spec that describes this work.
Code follows the module structure standards.
Wrote unit and integration tests
Updated relevant documentation (docs/) or specification (x/<module>/spec/)
Added relevant godoc comments.
Added a relevant changelog entry to the Unreleased section in CHANGELOG.md
Re-reviewed Files changed in the Github PR explorer
Review Codecov Report in the comment section below once CI passes

baseapp/msg_service_router.go

fdymylja · 2021-04-20T09:09:24Z

@AmauryM just some small feedback, I'd create a function in types:

func MsgName(msg Msg) { return proto.MessageName(msg) }

Since the message name has become important and we might see it more often, I think it would be better to have this functionality self contained in the types package rather than importing proto whenever the name of a Msg is needed.

codecov · 2021-04-20T14:27:23Z

Codecov Report

Merging #9139 (e95ae32) into master (6fbded9) will increase coverage by 0.05%.
The diff coverage is 64.61%.

@@            Coverage Diff             @@
##           master    #9139      +/-   ##
==========================================
+ Coverage   60.06%   60.11%   +0.05%     
==========================================
  Files         595      595              
  Lines       37355    37184     -171     
==========================================
- Hits        22437    22355      -82     
+ Misses      12929    12848      -81     
+ Partials     1989     1981       -8

Impacted Files	Coverage Δ
codec/types/interface_registry.go	`57.95% <ø> (-4.55%)`	⬇️
server/rosetta/client_online.go	`0.00% <0.00%> (ø)`
types/codec.go	`66.66% <ø> (+16.66%)`	⬆️
types/simulation/types.go	`0.00% <0.00%> (ø)`
types/tx_msg.go	`0.00% <0.00%> (ø)`
x/authz/keeper/msg_server.go	`0.00% <0.00%> (ø)`
x/authz/types/codec.go	`0.00% <0.00%> (ø)`
x/authz/types/generic_authorization.go	`60.00% <ø> (-11.43%)`	⬇️
x/bank/client/testutil/cli_helpers.go	`50.00% <ø> (+41.89%)`	⬆️
x/feegrant/types/filtered_fee.go	`0.00% <0.00%> (ø)`
... and 29 more

x/authz/simulation/operations.go

robert-zaremba

Great simplification and great work!
Left few comments.

CHANGELOG.md

baseapp/baseapp.go

baseapp/msg_service_router.go

robert-zaremba · 2021-04-22T09:40:33Z

baseapp/msg_service_router.go

+				// We panic here because there is no other alternative and the app cannot be initialized correctly
+				// this should only happen if there is a problem with code generation in which case the app won't
+				// work correctly anyway.
+				panic(fmt.Errorf("can't register request type %T for service method %s", i, fqMethod))


why we don't return an error?

If we return an error, then the app will continue running, with potential encoding errors that are hard to debug down the road.

robert-zaremba · 2021-04-22T09:40:59Z

baseapp/msg_service_router.go

+		// NOTE: This is how we pull the concrete request type for each handler for registering in the InterfaceRegistry.
+		// This approach is maybe a bit hacky, but less hacky than reflecting on the handler object itself.
+		// We use a no-op interceptor to avoid actually calling into the handler itself.
+		_, _ = methodHandler(nil, context.Background(), func(i interface{}) error {


is there context available in sd? We should use it instead of context.Background

Unfortunately not, sd is just a descriptor.

robert-zaremba · 2021-04-29T13:54:19Z

types/tx_msg.go

 		// ValidateBasic does a simple validation check that
 		// doesn't require access to any other information.
 		ValidateBasic() error

-		// Get the canonical byte representation of the Msg.
-		GetSignBytes() []byte
-


Shall we update the code generator as well?

I'm not sure which code generator you're talking about. The one in regen?

I thought these functions are generated (didn't check it).

robert-zaremba · 2021-04-29T13:57:08Z

x/auth/legacy/legacytx/stdtx_builder.go

-		}
-	}
-	s.Msgs = stdTxMsgs
+	s.Msgs = msgs
 	return nil
 }


shall we update the interface to not return an error?

let's update TxBuilder once, when we decide on a good TxBuilder DX (#8138)

x/authz/client/rest/grpc_query_test.go

robert-zaremba · 2021-04-29T14:02:41Z

x/bank/types/send_authorization.go

@@ -21,26 +19,25 @@ func NewSendAuthorization(spendLimit sdk.Coins) *SendAuthorization {

 // MethodName implements Authorization.MethodName.
 func (authorization SendAuthorization) MethodName() string {
-	return "/cosmos.bank.v1beta1.Msg/Send"
+	return sdk.MsgTypeURL(&MsgSend{})


let's move it to a private (module) var. No need to recreate it. We can update other places as well, where we have similar functions.

I tried this, but it actually doesn't work. proto.MessageName(&MsgSend{}) == "" on the module level. I think it's related to the ordering of which modules are run first. I'm not 100% sure how to make sure proto registration is run before defining a module var in this file.

See for example this failing test: https://github.com/cosmos/cosmos-sdk/pull/9139/checks?check_run_id=2467765703#step:6:155

you can import like this:

import _ "gitrhub.com/cosmos-sdk/x/bank/types"

this will cause the proto init function in bank generated pb files to be run and the msg name to be correctly resolved.

Good point, thanks!

Which do you think is clearer? Adding import _, or just recreating a sdk.MsgTypeURL() in each function? I somehow feel the latter is better for readability, and less hidden knowledge.

Maybe RegisterInterfaces can set these values?

robert-zaremba · 2021-04-29T14:03:09Z

x/evidence/types/msgs_test.go

-		require.Equal(t, tc.msg.Route(), types.RouterKey, "unexpected result for tc #%d", i)
-		require.Equal(t, tc.msg.Type(), types.TypeMsgSubmitEvidence, "unexpected result for tc #%d", i)
+		require.Equal(t, tc.msg.(legacytx.LegacyMsg).Route(), types.RouterKey, "unexpected result for tc #%d", i)
+		require.Equal(t, tc.msg.(legacytx.LegacyMsg).Type(), types.TypeMsgSubmitEvidence, "unexpected result for tc #%d", i)


why this is LegacyMsg?

We moved sdk.Msg to legacytx.LegacyMsg. So in the SDK all msgs are actually stilllegacytx.LegacyMsg.

…dr-31-type-url

amaury1093 · 2021-04-29T14:52:49Z

@robert-zaremba I addressed all reviews that made sense to me, left comments for the others.

technicallyty

LGTM 🚀 Good stuff - msgs look much simpler/cleaner with this PR!

x/authz/keeper/keeper_test.go

…dr-31-type-url

amaury1093 · 2021-04-30T11:00:18Z

@robert-zaremba I'm putting automerge on to unblock your work on Authz. I'll also work on a follow-up PR cleaning stuff (#9139 (comment), #9139 (comment), #9139 (comment)), if there are other requested changes we can address them there.

edit; #9236

robert-zaremba · 2021-04-30T13:05:50Z

@robert-zaremba I'm putting automerge on to unblock your work on Authz. I'll also work on a follow-up PR cleaning stuff (#9139 (comment), #9139 (comment), #9139 (comment)), if there are other requested changes we can address them there.

edit; #9236

Thanks, it's good.

colin-axner · 2021-05-07T11:29:08Z

baseapp/baseapp.go

 		)

-		if svcMsg, ok := msg.(sdk.ServiceMsg); ok {
-			msgFqName = svcMsg.MethodName


The message.action is very important for relayers as it is used to distinguish when the next message event starts/ends and what type of event is occurring. Relayers may be connected to chains on the old version and the new version and thus need to be aware of this change to ensure IBC continues to work between chains with different versions

Does the svcMsg.MethodName != sdk.MsgTypeURL(msg)? Or is there no breaking change here for Msgs migrated to proto messages registered on a Msg service (which from my understanding is the case for the current IBC Msgs). I'm a little fuzzy on the exact differences between the legacy messages, the proto message req/resp types and the msg service router, so please excuse me if I used the terminology wrong

It's just confusing, to me too. I tried to simplify it like this:

In v0.42, these message.actions were emitted:

either message.action='/cosmos.bank.v1beta1.Msg/Send' if the user explicity sent a ServiceMsg tx (note the 2 /)

or message.action='send' for all other messages (99% of the messages)

After this PR:

(1) is completely removed

(2) has been converted to message.action='/cosmos.bank.v1beta1.MsgSend' (note only 1 /), so legacy message.action='send' are not emitted anymore (unless the app developer didn't wire up proto registration correctly). This is not retroactive though.

Let me know if that works for IBC.

I see, thanks for the response! This makes sense. This should be fine for relayers, but it will likely require a small change. Relayer should now check for message.action='send' || message.action='/cosmos.bank.v1beta1.MsgSend'

Yes, that's the approach I'm taking too, e.g. if querying deposits here.

colin-axner · 2021-05-07T11:59:52Z

baseapp/baseapp.go

 			// legacy sdk.Msg routing
-			msgRoute := msg.Route()
-			msgFqName = msg.Type()
+			// Assuming that the app developer has migrated all their Msgs to
+			// proto messages and has registered all `Msg services`, then this
+			// path should never be called, because all those Msgs should be
+			// registered within the `msgServiceRouter` already.
+			msgRoute := legacyMsg.Route()
+			eventMsgName = legacyMsg.Type()


Also a sanity check, ledger signed transactions would not end up here if there is a registered proto message? This would be because the difference between the transaction/message layer?

For the modules in the SDK is this ever reached? Docs indicate that the handler type will be deprecated, but it seems like now all the SDK modules could remove their legacy Handler if it is no longer used? Just trying to get a better grasp of what is occurring

For the modules in the SDK is this ever reached?

For SDK modules, this is never reached. I just confirmed by adding a panic in this else if branch, all simapp/integration/CLI tests pass (only some baseapp unit tests panic, because these tests don't register proto msgs).

but it seems like now all the SDK modules could remove their legacy Handler if it is no longer used?

That's also correct, but we're thinking a better AppModule in general, so it might make sense to remove the legacy Handler with that in one go. ref: #9139 (comment)

Based on discussion from cosmos/cosmos-sdk#9139 (comment) This is not yet tested!

* Adapted proto compiler. Regenerated Rust files * Adapt to newer namespace for apps. * Manual fix for Staking erroneous compilation * Workaround: disabled chain upgrade * Added specific GRPC-web port to prevent conflict on 9091. * Upgrade protos * Add support for new message action strings * Improved output for one-chain script * CLI for chain upgrade proposal adapted to gaia v5. * Re-generated proto files with updated proto-compiler (fmt enabled). * Manual fix for staking::Validators erroneous compilation; added the .applications.transfer generated file * Quick feature for parametrizing upgraded chain * Clippy fix. Meta for params for better display. * Adapted query_upgraded_client_state * Adapted query_upgraded_consensus_state to ibc-go-v1 * Moved from gRPC to ABCI * Adapted send_tx_simulate to new def of SimulateRequest * Added parametrizable upgrade plan name * Fix unwraps. Retain unbonding period if none specified. * Missing post-merge bracket * Removed register and send match arms for rpc events * Support for new message.actions strings of ICS27 Based on discussion from cosmos/cosmos-sdk#9139 (comment) This is not yet tested! * Update protos sdk(v0.43.0-rc2) & ibc-go(v1.0.0-rc3) * Revert "Added specific GRPC-web port to prevent conflict on 9091." This reverts commit abf6dba. * Add missing fields in upgrade Plan * Fix clippy warnings * Added another action field to match intertx MsgSend * Update protos sdk(v0.43.0) & ibc-go(v1.0.0) * Update cosmos SDK module version requirement in compatibility.rs * Add legacy upgrade support (#1289) * Add legacy upgrade support * Fix check for legacy * Update .changelog * Add TODO * Apply suggestion Co-authored-by: Adi Seredinschi <adi@informal.systems> Co-authored-by: Adi Seredinschi <adi@informal.systems> * Apply suggestions * Update .changelog * Fix .changelog entry Co-authored-by: Adi Seredinschi <adi@informal.systems> Co-authored-by: Shoaib Ahmed <sufialhussaini@gmail.com>

* Adapted proto compiler. Regenerated Rust files * Adapt to newer namespace for apps. * Manual fix for Staking erroneous compilation * Workaround: disabled chain upgrade * Added specific GRPC-web port to prevent conflict on 9091. * Upgrade protos * Add support for new message action strings * Improved output for one-chain script * CLI for chain upgrade proposal adapted to gaia v5. * Re-generated proto files with updated proto-compiler (fmt enabled). * Manual fix for staking::Validators erroneous compilation; added the .applications.transfer generated file * Quick feature for parametrizing upgraded chain * Clippy fix. Meta for params for better display. * Adapted query_upgraded_client_state * Adapted query_upgraded_consensus_state to ibc-go-v1 * Moved from gRPC to ABCI * Adapted send_tx_simulate to new def of SimulateRequest * Added parametrizable upgrade plan name * Fix unwraps. Retain unbonding period if none specified. * Missing post-merge bracket * Removed register and send match arms for rpc events * Support for new message.actions strings of ICS27 Based on discussion from cosmos/cosmos-sdk#9139 (comment) This is not yet tested! * Update protos sdk(v0.43.0-rc2) & ibc-go(v1.0.0-rc3) * Revert "Added specific GRPC-web port to prevent conflict on 9091." This reverts commit abf6dba. * Add missing fields in upgrade Plan * Fix clippy warnings * Added another action field to match intertx MsgSend * Update protos sdk(v0.43.0) & ibc-go(v1.0.0) * Update cosmos SDK module version requirement in compatibility.rs * Add legacy upgrade support (informalsystems#1289) * Add legacy upgrade support * Fix check for legacy * Update .changelog * Add TODO * Apply suggestion Co-authored-by: Adi Seredinschi <adi@informal.systems> Co-authored-by: Adi Seredinschi <adi@informal.systems> * Apply suggestions * Update .changelog * Fix .changelog entry Co-authored-by: Adi Seredinschi <adi@informal.systems> Co-authored-by: Shoaib Ahmed <sufialhussaini@gmail.com>

* Adapted proto compiler. Regenerated Rust files * Adapt to newer namespace for apps. * Manual fix for Staking erroneous compilation * Workaround: disabled chain upgrade * Added specific GRPC-web port to prevent conflict on 9091. * Upgrade protos * Add support for new message action strings * Improved output for one-chain script * CLI for chain upgrade proposal adapted to gaia v5. * Re-generated proto files with updated proto-compiler (fmt enabled). * Manual fix for staking::Validators erroneous compilation; added the .applications.transfer generated file * Quick feature for parametrizing upgraded chain * Clippy fix. Meta for params for better display. * Adapted query_upgraded_client_state * Adapted query_upgraded_consensus_state to ibc-go-v1 * Moved from gRPC to ABCI * Adapted send_tx_simulate to new def of SimulateRequest * Added parametrizable upgrade plan name * Fix unwraps. Retain unbonding period if none specified. * Missing post-merge bracket * Removed register and send match arms for rpc events * Support for new message.actions strings of ICS27 Based on discussion from cosmos/cosmos-sdk#9139 (comment) This is not yet tested! * Update protos sdk(v0.43.0-rc2) & ibc-go(v1.0.0-rc3) * Revert "Added specific GRPC-web port to prevent conflict on 9091." This reverts commit abf6dba19f8eb5fcc370fcb33edc9bf5a1c4c21a. * Add missing fields in upgrade Plan * Fix clippy warnings * Added another action field to match intertx MsgSend * Update protos sdk(v0.43.0) & ibc-go(v1.0.0) * Update cosmos SDK module version requirement in compatibility.rs * Add legacy upgrade support (#1289) * Add legacy upgrade support * Fix check for legacy * Update .changelog * Add TODO * Apply suggestion Co-authored-by: Adi Seredinschi <adi@informal.systems> Co-authored-by: Adi Seredinschi <adi@informal.systems> * Apply suggestions * Update .changelog * Fix .changelog entry Co-authored-by: Adi Seredinschi <adi@informal.systems> Co-authored-by: Shoaib Ahmed <sufialhussaini@gmail.com>

aaronc and others added 7 commits April 12, 2021 17:25

wip

f1398cb

wip

b54d71f

wip

03a358d

wip on refactoring adr 031 type URLs

2a6223e

Fix msg_service_router

559fc54

Fix gov client queries

f2d8355

Fix some modules tests

d5c6ce9

amaury1093 commented Apr 19, 2021

View reviewed changes

baseapp/msg_service_router.go Outdated Show resolved Hide resolved

amaury1093 added 8 commits April 19, 2021 12:54

Remove all instances of "*.Msg/*"

569f777

Merge branch 'master' into am/adr-31-type-url

19720e7

Uncomment code

c52587c

Remove commented code

aa30868

// simulation.NewWeightedOperation(

2919375

Fix CopyTx

488dc62

Fix more tests

72c0cbc

Fix x/gov test

04ba707

amaury1093 added 7 commits April 20, 2021 11:43

proto.MessageName->sdk.MsgName

5f543a4

Fix authz tests

6d1b1cc

Merge branch 'master' into am/adr-31-type-url

16b72f1

Use MsgRoute in feegrant and staking/authz

d4130c2

Fix more tests

5f4a19c

Fix sims?

d5a2982

Add norace tag

9743b9e

Add CL

a323bff

amaury1093 commented Apr 20, 2021

View reviewed changes

x/authz/simulation/operations.go Outdated Show resolved Hide resolved

amaury1093 marked this pull request as ready for review April 20, 2021 14:44

amaury1093 requested review from alessio and alexanderbez as code owners April 20, 2021 14:44

github-actions bot added T: ADR An issue or PR relating to an architectural decision record T:Docs Changes and features related to documentation. labels Apr 28, 2021

robert-zaremba reviewed Apr 29, 2021

View reviewed changes

Merge branch 'master' of ssh://github.com/cosmos/cosmos-sdk into am/a…

d2a8ae5

…dr-31-type-url

Address reviews

cd4bc02

amaury1093 force-pushed the am/adr-31-type-url branch from 4f2c8e4 to cd4bc02 Compare April 29, 2021 15:22

Fix test

774a4ab

technicallyty approved these changes Apr 29, 2021

View reviewed changes

x/authz/keeper/keeper_test.go Outdated Show resolved Hide resolved

x/authz/keeper/keeper_test.go Outdated Show resolved Hide resolved

amaury1093 added 3 commits April 30, 2021 12:38

Remove stray log

d3c3381

Merge branch 'master' of ssh://github.com/cosmos/cosmos-sdk into am/a…

98ad2b7

…dr-31-type-url

Update CL

e95ae32

amaury1093 added the A:automerge Automatically merge PR once all prerequisites pass. label Apr 30, 2021

mergify bot merged commit dfe3e7a into master Apr 30, 2021

mergify bot deleted the am/adr-31-type-url branch April 30, 2021 11:00

amaury1093 mentioned this pull request Apr 30, 2021

Cleanup remains of ServiceMsg #9236

Merged

9 tasks

colin-axner reviewed May 7, 2021

View reviewed changes

colin-axner mentioned this pull request May 7, 2021

Event parsing message.action preparation for new SDK version informalsystems/hermes#918

Closed

5 tasks

amaury1093 mentioned this pull request May 10, 2021

x/staking v0.43 Audit updates #9267

Merged

9 tasks

cgorenflo mentioned this pull request May 11, 2021

Replace handlers with Msg Services axelarnetwork/axelar-core#448

Closed

7 tasks

blushi mentioned this pull request May 28, 2021

Manual Testing for v0.43 regen-network/regen-ledger#352

Closed

56 tasks

adizere added a commit to informalsystems/hermes that referenced this pull request Jul 15, 2021

Support for new message.actions strings of ICS27

b912347

Based on discussion from cosmos/cosmos-sdk#9139 (comment) This is not yet tested!

dadamu mentioned this pull request Jul 23, 2021

Remove all the legacy handlers desmos-labs/desmos#569

Closed

7 tasks

technicallyty mentioned this pull request Aug 23, 2021

[codec] Calling proto.MessageName on sdk.Msgs after decoding a TX returns an empty string #8728

Closed

4 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Remove ServiceMsgs from ADR-031 #9139

Remove ServiceMsgs from ADR-031 #9139

amaury1093 commented Apr 19, 2021 •

edited

Loading

fdymylja commented Apr 20, 2021

codecov bot commented Apr 20, 2021 •

edited

Loading

robert-zaremba left a comment

robert-zaremba Apr 22, 2021

amaury1093 Apr 29, 2021 •

edited

Loading

robert-zaremba Apr 22, 2021

amaury1093 Apr 29, 2021

robert-zaremba Apr 29, 2021

robert-zaremba Apr 29, 2021

amaury1093 Apr 29, 2021

robert-zaremba Apr 30, 2021

robert-zaremba Apr 29, 2021

amaury1093 Apr 29, 2021

robert-zaremba Apr 29, 2021

amaury1093 Apr 29, 2021 •

edited

Loading

fdymylja Apr 30, 2021 •

edited

Loading

amaury1093 Apr 30, 2021

robert-zaremba Apr 30, 2021

robert-zaremba Apr 29, 2021

amaury1093 Apr 29, 2021 •

edited

Loading

amaury1093 commented Apr 29, 2021

technicallyty left a comment

amaury1093 commented Apr 30, 2021 •

edited

Loading

robert-zaremba commented Apr 30, 2021

colin-axner May 7, 2021

amaury1093 May 7, 2021 •

edited

Loading

colin-axner May 7, 2021

amaury1093 May 7, 2021

colin-axner May 7, 2021

colin-axner May 7, 2021

amaury1093 May 7, 2021 •

edited

Loading

Remove ServiceMsgs from ADR-031 #9139

Remove ServiceMsgs from ADR-031 #9139

Conversation

amaury1093 commented Apr 19, 2021 • edited Loading

Description

fdymylja commented Apr 20, 2021

codecov bot commented Apr 20, 2021 • edited Loading

Codecov Report

robert-zaremba left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

amaury1093 Apr 29, 2021 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

amaury1093 Apr 29, 2021 • edited Loading

Choose a reason for hiding this comment

fdymylja Apr 30, 2021 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

amaury1093 Apr 29, 2021 • edited Loading

Choose a reason for hiding this comment

amaury1093 commented Apr 29, 2021

technicallyty left a comment

Choose a reason for hiding this comment

amaury1093 commented Apr 30, 2021 • edited Loading

robert-zaremba commented Apr 30, 2021

Choose a reason for hiding this comment

amaury1093 May 7, 2021 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

amaury1093 May 7, 2021 • edited Loading

Choose a reason for hiding this comment

amaury1093 commented Apr 19, 2021 •

edited

Loading

codecov bot commented Apr 20, 2021 •

edited

Loading

amaury1093 Apr 29, 2021 •

edited

Loading

amaury1093 Apr 29, 2021 •

edited

Loading

fdymylja Apr 30, 2021 •

edited

Loading

amaury1093 Apr 29, 2021 •

edited

Loading

amaury1093 commented Apr 30, 2021 •

edited

Loading

amaury1093 May 7, 2021 •

edited

Loading

amaury1093 May 7, 2021 •

edited

Loading