Add experimental support for MSC3768: Push rule action for in-app notifications #18720

Johennes · 2025-07-23T13:53:52Z

This adds experimental support for matrix-org/matrix-spec-proposals#3768.

Pull Request Checklist

Pull request is based on the develop branch
Pull request includes a changelog file. The entry should:
- Be a short description of your change which makes sense to users. "Fixed a bug that prevented receiving messages from other servers." instead of "Moved X method from EventStore to EventWorkerStore.".
- Use markdown where necessary, mostly for code blocks.
- End with either a period (.) or an exclamation mark (!).
- Start with a capital letter.
- Feel free to credit yourself, by adding a sentence "Contributed by @github_username." or "Contributed by [Your Name]." to the end of the entry.
Code style is correct (run the linters)

synapse/handlers/push_rules.py

…ifications Signed-off-by: Johannes Marbach <[email protected]>

synapse/handlers/push_rules.py

MadLittleMods · 2025-07-24T16:52:02Z

synapse/handlers/push_rules.py

@@ -154,6 +154,9 @@ def check_actions(actions: List[Union[str, JsonDict]]) -> None:
        # ignored (resulting in no action from the pusher).
        if a in ["notify", "dont_notify", "coalesce"]:
            pass
+        # In-app only notification as per MSC3768
+        elif a == "org.matrix.msc3768.notify_in_app":


"org.matrix.msc3768.notify_in_app" should be a constant; along with the others but we should at-least do good with this new one for this PR.

I've added one for org.matrix.msc3768.notify_in_app in 4215344. I can also add constants for the other three if you're ok with the PR growing a bit? Looks like it'd be 30 to 40 occurrences to replace.

I can also add constants for the other three if you're ok with the PR growing a bit? Looks like it'd be 30 to 40 occurrences to replace.

I'd be fine with it but for the sake of others trying to follow along on this PR we should do it separately if you're keen.

tests/push/test_http.py

tests/replication/storage/test_events.py

changelog.d/18720.feature

MadLittleMods · 2025-07-24T17:00:47Z

rust/src/push/mod.rs

@@ -174,6 +174,9 @@ pub enum Action {
    Notify,
    SetTweak(SetTweak),

+    // In-app only notification as per MSC3768
+    NotifyInApp,


Do we need to handle this at this spot given "org.matrix.msc3768.notify_in_app" shouldn't result in push notifications?

synapse/rust/src/push/evaluator.rs

Lines 227 to 229 in 458e641

// Filter out "dont_notify" and "coalesce" actions, as we don't store them

// (since they result in no action by the pushers).

.filter(|a| **a != Action::DontNotify && **a != Action::Coalesce)

Hm, I think this should stay as it is. The doc comment for run says:

/// Returns the set of actions, if any, that match (filtering out any
/// dont_notify and coalesce actions).

dont_notify and coalesce are probably removed because they are literally equivalent to an empty actions array. If we also remove notify_in_app here, I think that would lead to the event not causing any notification at all?

Co-authored-by: Eric Eastwood <[email protected]>

Signed-off-by: Johannes Marbach <[email protected]>

Co-authored-by: Eric Eastwood <[email protected]>

Signed-off-by: Johannes Marbach <[email protected]>

Co-authored-by: Eric Eastwood <[email protected]>

Signed-off-by: Johannes Marbach <[email protected]>

MadLittleMods · 2025-07-28T15:38:56Z

changelog.d/18720.feature

@@ -0,0 +1 @@
+Add experimental support for MSC3768 (push rule action for in-app notifications).


Given this discussion on the MSC that this might be better suited as a responsibility of the client or OS, should this PR/MSC move forward? (seems like you agree with the points)

I was thinking that suppressing pushes on the server could be a MAY. It's relatively simple and saves at least some processing cost. So generally, I'd still like this to go ahead. However, for the sake of the MSC, we could also leave it in draft if you'd rather not merge it. Draft PRs are usually sufficient for the process, especially when they received reviewed.

If leaving the PR unmerged is viable to you, that works for me. I'm not really convinced by the concept at the moment.

You mean on the Synapse side or in general?

Based on the current reasoning in the MSC, I'm not seeing the value of this proposal yet. The Synapse implementation appears technically sound, but I'm open to either:

Not merging this PR if you agree, or

Gathering additional perspectives to help us decide.

Ok, I see. No, I'm totally fine with not merging this. Just thought you might have feedback on the MSC.

Moving this back to draft since it seems like functionally we're done. Thanks a lot for your review. 🙏

synapse/api/constants.py

tests/replication/storage/test_events.py

Co-authored-by: Eric Eastwood <[email protected]>

synapse/handlers/push_rules.py

MadLittleMods · 2025-07-29T14:39:45Z

changelog.d/18720.feature

@@ -0,0 +1 @@
+Add experimental support for MSC3768 (push rule action for in-app notifications).


If leaving the PR unmerged is viable to you, that works for me. I'm not really convinced by the concept at the moment.

MadLittleMods · 2025-07-29T14:56:06Z

synapse/handlers/push_rules.py

+            else:
+                pass  # The spec doesn't forbid introducing custom actions


Previous discussion, #18720 (comment)

If we're doing this behavior and can justify it, we should just do it all the time.

This would basically make the whole function obsolete then?

Seems like mostly. I think we could still validate known rules like set_tweak 🤷 although an unknown variant should also pass by according to the spec.

Yeah. 🤔

Well given that we won't merge this at the moment, I'm going to leave this unaddressed for now.

tests/replication/storage/test_events.py

Signed-off-by: Johannes Marbach <[email protected]>

Johennes force-pushed the johannes/msc3768-notify-in-app branch 2 times, most recently from 848e221 to 5f921e4 Compare July 23, 2025 13:55

Johennes commented Jul 23, 2025

View reviewed changes

synapse/handlers/push_rules.py Outdated Show resolved Hide resolved

Johennes force-pushed the johannes/msc3768-notify-in-app branch from 5f921e4 to 948514a Compare July 24, 2025 09:09

Johennes marked this pull request as ready for review July 24, 2025 09:44

Johennes requested a review from a team as a code owner July 24, 2025 09:44

Add experimental support for MSC3768: Push rule action for in-app not…

389a870

…ifications Signed-off-by: Johannes Marbach <[email protected]>

Johennes force-pushed the johannes/msc3768-notify-in-app branch from 948514a to 389a870 Compare July 24, 2025 10:20

MadLittleMods added the A-Push label Jul 24, 2025

MadLittleMods reviewed Jul 24, 2025

View reviewed changes

Johennes and others added 8 commits July 25, 2025 13:50

Update tests/push/test_http.py

2d9ebf5

Co-authored-by: Eric Eastwood <[email protected]>

Fix comment

72d1a31

Signed-off-by: Johannes Marbach <[email protected]>

Update tests/push/test_http.py

0b353ed

Co-authored-by: Eric Eastwood <[email protected]>

Use EventTypes constants

6fcccba

Signed-off-by: Johannes Marbach <[email protected]>

Add missing comments

6ab6f3c

Signed-off-by: Johannes Marbach <[email protected]>

Update changelog.d/18720.feature

34ae3c3

Co-authored-by: Eric Eastwood <[email protected]>

Extract a constant for org.matrix.msc3768.notify_in_app

4215344

Signed-off-by: Johannes Marbach <[email protected]>

Merge branch 'develop' into johannes/msc3768-notify-in-app

97d50cd

Johennes requested a review from MadLittleMods July 25, 2025 14:27

Johennes added 2 commits July 28, 2025 13:42

Update check_actions

6ee3977

Signed-off-by: Johannes Marbach <[email protected]>

Merge branch 'develop' into johannes/msc3768-notify-in-app

632b3b4

MadLittleMods reviewed Jul 28, 2025

View reviewed changes

Johennes and others added 4 commits July 29, 2025 15:55

Update tests/replication/storage/test_events.py

24e06df

Co-authored-by: Eric Eastwood <[email protected]>

Add missing import and replace further m.text usages

b2b3586

MSC_3768_NOTIFY_IN_APP -> MSC3768_NOTIFY_IN_APP

a2f9789

Split test

5633084

MadLittleMods reviewed Jul 29, 2025

View reviewed changes

Johennes added 2 commits July 30, 2025 14:58

Add missing doc string for config argument

f1ab6ac

Signed-off-by: Johannes Marbach <[email protected]>

Replace membership literals with constants

816a7ff

Signed-off-by: Johannes Marbach <[email protected]>

Johennes marked this pull request as draft July 30, 2025 13:06

Johennes mentioned this pull request Jul 30, 2025

MSC3768: Push rule action for in-app notifications matrix-org/matrix-spec-proposals#3768

Open

	// Filter out "dont_notify" and "coalesce" actions, as we don't store them
	// (since they result in no action by the pushers).
	.filter(\|a\| a != Action::DontNotify && a != Action::Coalesce)

		@@ -0,0 +1 @@
		Add experimental support for MSC3768 (push rule action for in-app notifications).

		else:
		pass # The spec doesn't forbid introducing custom actions

Add experimental support for MSC3768: Push rule action for in-app notifications #18720

Are you sure you want to change the base?

Add experimental support for MSC3768: Push rule action for in-app notifications #18720

Uh oh!

Conversation

Johennes commented Jul 23, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Pull Request Checklist

Uh oh!

Uh oh!

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!

Uh oh!

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!

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!

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!

Choose a reason for hiding this comment

Uh oh!

MadLittleMods Jul 30, 2025 • 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!

Uh oh!

Uh oh!

Johennes commented Jul 23, 2025 •

edited

Loading

MadLittleMods Jul 30, 2025 •

edited

Loading