feat: Enable components to automatically accept multiple connections if their input type is `list` #10441

sjrl · 2026-01-23T13:55:03Z

Related Issues

fixes https://github.com/deepset-ai/haystack-private/issues/241

Proposed Changes:

Natively support connecting multiple outputs directly to another component without needing the user to define a Joiner component.

For example, a user may want to connect multiple output edges of the FileTypeRouter to all go to the same converter (e.g. ImageConverter). Before you would have to define a ListJoiner as an in-between component. The changes introduced in this PR would allow you to directly connect all outputs from the FileTypeRouter to the ImageConverter.

Here is an example

from haystack import Pipeline
from haystack.components.converters import HTMLToDocument, TextFileToDocument
from haystack.components.routers import FileTypeRouter
from haystack.components.writers import DocumentWriter
from haystack.dataclasses import ByteStream
from haystack.document_stores.in_memory import InMemoryDocumentStore

sources = [
    ByteStream.from_string(text="Text file content", mime_type="text/plain", meta={"file_type": "txt"}),
    ByteStream.from_string(
        text="\n<html><body>Some content</body></html>\n", mime_type="text/html", meta={"file_type": "html"},
    ),
]

doc_store = InMemoryDocumentStore()
pipe = Pipeline()

pipe.add_component("router", FileTypeRouter(mime_types=["text/plain", "text/html"]))
pipe.add_component("txt_converter", TextFileToDocument())
pipe.add_component("html_converter", HTMLToDocument())
pipe.add_component("writer", DocumentWriter(doc_store))

pipe.connect("router.text/plain", "txt_converter.sources")
pipe.connect("router.text/html", "html_converter.sources")
# The DocumentWriter accepts documents from both converters without needing a DocumentJoiner
pipe.connect("txt_converter.documents", "writer.documents")
pipe.connect("html_converter.documents", "writer.documents")

result = pipe.run({"router": {"sources": sources}})
# result["writer"]["documents_written"] == 2

How did you test it?

Added two new pipeline run tests

Notes for the reviewer

Checklist

I have read the contributors guidelines and the code of conduct.
I have updated the related issue with new insights and changes.
I have added unit tests and updated the docstrings.
I've used one of the conventional commit types for my PR title: fix:, feat:, build:, chore:, ci:, docs:, style:, refactor:, perf:, test: and added ! in case the PR includes breaking changes.
I have documented my code.
I have added a release note file, following the contributors guidelines.
I have run pre-commit hooks and fixed any issue.

vercel · 2026-01-23T13:55:09Z

The latest updates on your projects. Learn more about Vercel for GitHub.

1 Skipped Deployment

Project	Deployment	Review	Updated (UTC)
haystack-docs	Ignored	Preview	Jan 28, 2026 1:16pm

…dic-input

coveralls · 2026-01-23T14:01:18Z

Pull Request Test Coverage Report for Build 21361994530

Warning: This coverage report may be inaccurate.

This pull request's base commit is no longer the HEAD commit of its target branch. This means it includes changes from outside the original pull request, including, potentially, unrelated coverage changes.

For more information on this, see Tracking coverage changes with pull request builds.
To avoid this issue with future PRs, see these Recommended CI Configurations.
For a quick fix, rebase this PR at GitHub. Your next report should be accurate.

Details

0 of 0 changed or added relevant lines in 0 files are covered.
58 unchanged lines in 3 files lost coverage.
Overall coverage decreased (-0.02%) to 92.431%

Files with Coverage Reduction	New Missed Lines	%
core/component/types.py	3	92.68%
core/pipeline/pipeline.py	3	93.16%
core/pipeline/base.py	52	86.68%

Totals
Change from base Build 21361849247:	-0.02%
Covered Lines:	14763
Relevant Lines:	15972

💛 - Coveralls

…dic-input

haystack/core/pipeline/base.py

…dic-input

sjrl · 2026-01-28T08:11:03Z

@anakin87 you might also find this interesting to review given your work on the OutputAdapter issue

releasenotes/notes/add-auto-joiners-for-lists-types-b07771e6d8add3da.yaml

…add3da.yaml Co-authored-by: Stefano Fiorucci <stefanofiorucci@gmail.com>

sjrl · 2026-01-28T08:39:52Z

haystack/core/pipeline/base.py

+            # We automatically set the receiver socket as variadic if:
+            # - it has at least one sender already connected
+            # - it's not already variadic
+            # - its origin type is list
+            if not receiver_socket.is_variadic and _safe_get_origin(receiver_socket.type) == list:
+                receiver_socket.is_lazy_variadic = True
+                # We also disable wrapping inputs into list so the sender outputs matches the type of the receiver
+                # socket.
+                receiver_socket.wrap_input_in_list = False


I thought about making this it's own function so I could reuse it but it felt too short and unnecessarily obfusicated the logic of what was going on so I decided to leave as is

If we automatically set the receiver socket as variadic could that change the execution order of components in a pipeline in some edge cases? If we set it to variadic we need to postpone execution until we can be sure that there won't be any other inputs, right? Can we be sure that because of that change the component is not executed later than before ?

Yeah that should all be fine. I'm utilizing the same logic that already exists. Since the execution queue is not filled until the pipeline is run this receiver is just treated like a prexisting variadic component. I added a behavioural test that copied a complicated joiner scenario with branches that didn't always fire and it passes.

As for changing the order of this receiver component compared to before. I don't think that would really be an issue since before, a joiner component would have lived right before the receiver so it wouldn't execute until the joiner fired anyway.

haystack/core/pipeline/base.py

haystack/core/pipeline/component_checks.py

haystack/core/type_utils.py

test/core/pipeline/features/test_run.py

anakin87 · 2026-01-28T09:24:24Z

haystack/core/pipeline/base.py

@@ -580,14 +579,26 @@ def connect(self, sender: str, receiver: str) -> "PipelineBase":  # noqa: PLR091
            # This is already connected, nothing to do
            return self



Could you add some tests that cover changes in this file to test/core/pipeline/test_pipeline_base.py?

Added see this comment and this comment

Thank you!

If you merge main, Coveralls message should reflect the new coverage.

sjrl · 2026-01-28T09:28:48Z

haystack/core/pipeline/base.py

+                if socket.senders and socket_name in component_inputs:
+                    # We automatically set the receiver socket as lazy variadic if:
+                    # - it has at least one sender already connected
+                    # - it's not already variadic
+                    # - its origin type is list
+                    if not socket.is_variadic and _safe_get_origin(socket.type) == list:
+                        socket.is_lazy_variadic = True
+                        # We also disable wrapping inputs into list so the sender outputs matches the type of the
+                        # receiver socket.
+                        socket.wrap_input_in_list = False


This is where we make a component auto variadic if a user input is a second connection to a component.

anakin87 · 2026-01-28T10:59:50Z

I'd also explain this new feature in docs.

sjrl · 2026-01-28T11:02:27Z

I'd also explain this new feature in docs.

Good idea, do you know where in the docs this would ideally go?

anakin87 · 2026-01-28T11:07:35Z

I'd also explain this new feature in docs.

Good idea, do you know where in the docs this would ideally go?

I'm not sure. Somewhere in https://docs.haystack.deepset.ai/docs/pipelines or in child pages.
(I'd expect then to expand that section with features I'm planning to implement.)

sjrl · 2026-01-28T13:14:55Z

Writing it here so I don't forget, but a follow up idea would be to getting this to also work for components that have Any as their input type. E.g. ChatPromptBuilders. Where it would be cool that if both senders have the same type and the senders origin type are list then we could also do this auto lazy variadic update. An example I'm thinking of is sending two lists of docs from different sources to one input on a ChatPromptBuilder.

…into auto-variadic-input

sjrl · 2026-01-28T13:17:18Z

test/core/pipeline/test_pipeline_base.py

+class TestPipelineConnect:
+    def test_connect(self):
+        comp1 = component_class("Comp1", output_types={"value": int})()
+        comp2 = component_class("Comp2", input_types={"value": int})()
+        pipe = PipelineBase()
+        pipe.add_component("comp1", comp1)
+        pipe.add_component("comp2", comp2)
+        assert pipe.connect("comp1.value", "comp2.value") is pipe


I collected all connect tests under this class

sjrl · 2026-01-28T13:17:51Z

test/core/pipeline/test_pipeline_base.py

+    def test_connect_auto_variadic(self):
+        @component
+        class ListAcceptor:
+            @component.output_types(result=list[int])
+            def run(self, numbers: list[int]) -> dict[str, list[int]]:
+                return {"result": numbers}


Here is the new test I added to test the new auto-variadic behavior on pipeline connection

sjrl · 2026-01-28T13:18:18Z

test/core/pipeline/test_pipeline_base.py

+class TestValidateInput:
+    def test_validate_input_wrong_comp_name(self):
+        pipe = PipelineBase()
+        pipe.add_component("comp", FakeComponent())
+        with pytest.raises(ValueError, match="Component named 'wrong_comp_name' not found in the pipeline."):
+            pipe.validate_input(data={"wrong_comp_name": {"input_": "test"}})


This function looked to be completely untested at least from unit tests so I added quite few under this class

sjrl · 2026-01-28T13:19:50Z

I'd also explain this new feature in docs.

Good idea, do you know where in the docs this would ideally go?

I'm not sure. Somewhere in https://docs.haystack.deepset.ai/docs/pipelines or in child pages. (I'd expect then to expand that section with features I'm planning to implement.)

Is it fine if I do this in a separate PR? I can create a follow up issue

anakin87 · 2026-01-28T13:32:00Z

I'd also explain this new feature in docs.

Good idea, do you know where in the docs this would ideally go?

I'm not sure. Somewhere in https://docs.haystack.deepset.ai/docs/pipelines or in child pages. (I'd expect then to expand that section with features I'm planning to implement.)

Is it fine if I do this in a separate PR? I can create a follow up issue

Ok!

anakin87

Looks good.

I'd like @julian-risch to take a look too.

julian-risch

Looks good to me too. I just would like to briefly discuss if this PR can change the order of execution of different components. Would be an edge case. Or maybe you already thought about it?

anakin87

👍

sjrl added 2 commits January 22, 2026 13:06

Add notes for places that would need to be updated

33c5a65

First draft for supporting auto layz variadic

5715f3a

github-actions bot added topic:tests topic:core type:documentation Improvements on the docs labels Jan 23, 2026

Merge branch 'main' of github.com:deepset-ai/haystack into auto-varia…

df21693

…dic-input

vercel bot deployed to Preview January 23, 2026 13:58 View deployment

sjrl added 3 commits January 26, 2026 10:29

Merge branch 'main' of github.com:deepset-ai/haystack into auto-varia…

6e48ca7

…dic-input

Fix and add some dev comments

ee3a0a1

Add pipeline_run test using the automatic joining behavior

c528347

sjrl commented Jan 26, 2026

View reviewed changes

haystack/core/pipeline/base.py Outdated Show resolved Hide resolved

Merge branch 'main' of github.com:deepset-ai/haystack into auto-varia…

605c09a

…dic-input

vercel bot deployed to Preview January 26, 2026 14:52 View deployment

sjrl added 2 commits January 27, 2026 11:11

Finish todo

994e74f

Merge branch 'main' of github.com:deepset-ai/haystack into auto-varia…

2bb5775

…dic-input

vercel bot deployed to Preview January 27, 2026 10:15 View deployment

sjrl added 3 commits January 27, 2026 11:26

Fix mypy error

6b6d629

Add new test

90c0e4c

Add reno

21024c3

sjrl marked this pull request as ready for review January 28, 2026 08:10

sjrl requested a review from a team as a code owner January 28, 2026 08:10

sjrl requested review from julian-risch and removed request for a team January 28, 2026 08:10

update reno

40969c5

anakin87 reviewed Jan 28, 2026

View reviewed changes

releasenotes/notes/add-auto-joiners-for-lists-types-b07771e6d8add3da.yaml Outdated Show resolved Hide resolved

Update releasenotes/notes/add-auto-joiners-for-lists-types-b07771e6d8…

f1a2340

…add3da.yaml Co-authored-by: Stefano Fiorucci <stefanofiorucci@gmail.com>

sjrl commented Jan 28, 2026

View reviewed changes

haystack/core/pipeline/base.py Outdated Show resolved Hide resolved

Apply suggestion from @sjrl

dbc433a

sjrl commented Jan 28, 2026

View reviewed changes

haystack/core/pipeline/component_checks.py Show resolved Hide resolved

sjrl commented Jan 28, 2026

View reviewed changes

haystack/core/type_utils.py Show resolved Hide resolved

anakin87 reviewed Jan 28, 2026

View reviewed changes

test/core/pipeline/features/test_run.py Show resolved Hide resolved

anakin87 reviewed Jan 28, 2026

View reviewed changes

sjrl commented Jan 28, 2026

View reviewed changes

Expand dev comment

164be82

Add more unit tests and a bit of refactoring

d792e33

Merge branch 'auto-variadic-input' of github.com:deepset-ai/haystack …

0acfe1d

…into auto-variadic-input

sjrl commented Jan 28, 2026

View reviewed changes

sjrl requested a review from anakin87 January 28, 2026 13:19

sjrl mentioned this pull request Jan 28, 2026

Add docs for new auto joiner behavior of Pipelines #10462

Open

anakin87 reviewed Jan 28, 2026

View reviewed changes

sjrl mentioned this pull request Jan 28, 2026

Consider adding a de-duplication step to our rankers #10463

Open

julian-risch reviewed Jan 28, 2026

View reviewed changes

sjrl requested review from anakin87 and julian-risch January 29, 2026 10:26

anakin87 approved these changes Jan 29, 2026

View reviewed changes

julian-risch approved these changes Jan 29, 2026

View reviewed changes

sjrl merged commit 41c6404 into main Jan 29, 2026
22 checks passed

sjrl deleted the auto-variadic-input branch January 29, 2026 10:29

		@@ -580,14 +579,26 @@ def connect(self, sender: str, receiver: str) -> "PipelineBase": # noqa: PLR091
		# This is already connected, nothing to do
		return self

feat: Enable components to automatically accept multiple connections if their input type is list #10441

feat: Enable components to automatically accept multiple connections if their input type is list #10441

Uh oh!

Conversation

sjrl commented Jan 23, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Related Issues

Proposed Changes:

How did you test it?

Notes for the reviewer

Checklist

Uh oh!

vercel bot commented Jan 23, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

coveralls commented Jan 23, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Pull Request Test Coverage Report for Build 21361994530

Warning: This coverage report may be inaccurate.

Details

💛 - Coveralls

Uh oh!

Uh oh!

sjrl commented Jan 28, 2026

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!

sjrl Jan 28, 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!

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!

anakin87 commented Jan 28, 2026

Uh oh!

sjrl commented Jan 28, 2026

Uh oh!

anakin87 commented Jan 28, 2026

Uh oh!

sjrl commented Jan 28, 2026

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!

sjrl commented Jan 28, 2026

Uh oh!

anakin87 commented Jan 28, 2026

Uh oh!

anakin87 left a comment

Choose a reason for hiding this comment

Uh oh!

julian-risch left a comment

Choose a reason for hiding this comment

Uh oh!

anakin87 left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

feat: Enable components to automatically accept multiple connections if their input type is `list` #10441

feat: Enable components to automatically accept multiple connections if their input type is `list` #10441

sjrl commented Jan 23, 2026 •

edited

Loading

vercel bot commented Jan 23, 2026 •

edited

Loading

coveralls commented Jan 23, 2026 •

edited

Loading

sjrl Jan 28, 2026 •

edited

Loading