Feature/resolve timelines by maarten-ic · Pull Request #62 · multiscale/ymmsl-python

maarten-ic · 2026-07-16T12:22:46Z

Continuation of PR #27

Put determined timelines back on Component (absolute timelines) and Port (relative to Component?) objects
~~Update timeline annotations when importing/resolving (components are renamed to x.y.<component> and timeline annotations should as well to prevent clashes)~~
Move to v0_2.validation.timelines. Function: resolve_timelines() because it updates the timelines on the data objects
Cyclic dependency documentation (move to issue?)

Signature accepts a Sequence, but the logic previously only worked when providing a list. Other sequences silently didn't work.

Add `parent` property and `relative_to` (inspired by pathlib.Path)

maarten-ic · 2026-07-16T14:27:58Z

~~Update timeline annotations when importing/resolving (components are renamed to x.y.<component> and timeline annotations should as well to prevent clashes)~~

N.B. Resolving imports doesn't rename components (only models) and therefore shouldn't rename timelines.
~~Timelines should be renamed when flattening, but that logic resides in muscle3 so it's not part of this PR.~~

Edit: thinking a bit more about this, I'm not sure if we must rename timelines when flattening the hierarchical models:

Timelines that are exported on a model port must not change names
Timelines that are internal to a model will never interact with components outside the model and so this is not a problem for the logic in this PR if they have the same name

Include information on the conduits instead of just the components. Example error message: > Detected a dependency cycle in model 'cycle'. The component 'first' has an F_INIT port that depends on data produced by one of its own O_F or O_I ports: Conduit(first.final -> second.init) -> Conduit(second.final -> third.init) -> Conduit(third.final -> first.init). You may have an error in the conduits or may need a different coupling schema.

maarten-ic · 2026-07-17T10:16:55Z

N.B. I've improved the cyclic dependency error, hopefully this is clear enough without writing and pointing to a new documentation page.

…melines

LourensVeen · 2026-07-20T08:52:17Z

Edit: thinking a bit more about this, I'm not sure if we must rename timelines when flattening the hierarchical models:

Imagine a top-level model with two components a and b, with a dispatching to b. Both a and b are implemented using a submodel, and each of those submodels has a macro-micro model inside it with components named macro and micro, and with macro's F_INIT and O_F connected to model ports.

After flattening, we have a flat model with components a.macro, a.micro, b.macro and b.micro. Component a.micro is on timeline :a.macro and component b.micro is on timeline :b.macro. Without renaming during flattening, they'd both be on :macro and would appear to be on the same timeline while they're really not.

Timelines that are exported on a model port must not change names
Timelines that are internal to a model will never interact with components outside the model and so this is not a problem for the logic in this PR if they have the same name

Timelines on ports are usually relative timelines, and those shouldn't be a problem as long as the absolute timeline they're appended to is correct. They can interact with components outside the model just fine though, we can have a component macro inside of a submodel that has its O_I and S ports connected to O_I and S ports on the submodel, and a micro component in another submodel.

LourensVeen

Definitely headed in the right direction, nice work!

Based on PR feedback

maarten-ic · 2026-07-21T12:39:28Z

Edit: thinking a bit more about this, I'm not sure if we must rename timelines when flattening the hierarchical models:

Imagine a top-level model with two components a and b, with a dispatching to b. Both a and b are implemented using a submodel, and each of those submodels has a macro-micro model inside it with components named macro and micro, and with macro's F_INIT and O_F connected to model ports.

After flattening, we have a flat model with components a.macro, a.micro, b.macro and b.micro. Component a.micro is on timeline :a.macro and component b.micro is on timeline :b.macro. Without renaming during flattening, they'd both be on :macro and would appear to be on the same timeline while they're really not.

Yes, fully agreed! For complete correctness you'd have to rename these internal timelines.

My point (which I didn't describe clearly, sorry for that!) was that timeline renaming isn't required for the checks in this PR. The flattened model is more of an internal MUSCLE3 state and I expect it's not required to have the renaming functionality in to get a properly behaving simulation. We'll have to test though 🙂

LourensVeen · 2026-07-29T14:32:09Z

Ah, and after flattening the timelines aren't actually used anyway, so it would work even if we don't update them during flattening. Yes, I think that's right. And components don't get renamed during resolving, so there's no issue there either. All good.

LourensVeen

A few suggestions, but I think this is good to go?

LourensVeen · 2026-07-29T14:43:33Z

+    model. Any of the following exceptions (which are a subclass of
+    :class:`ResolveTimelineException`) will be raised when the model timelines are not
+    consistent. See their description for more details:
+


You can use a Raises: section for this similar to Args: to make Sphinx render it nicely:

Raises: :class:`CyclicDependency` if an O_F to F_INIT loop was found :class:`TooManyReducerFilters` if a timeline was reduced by more levels than it has

LourensVeen · 2026-07-29T14:43:53Z



-class CyclicDependency(RuntimeError):
+class ResolveTimelineException(RuntimeError):


LourensVeen · 2026-07-29T14:44:57Z

+    This error is raised when a reducer filter attempts to reduce a conduit that
+    already sends on the root timeline. :class:`InconsistentTimelines` or
+    :class:`ConduitTimelineError` may also be raised when too many reducer filters are
+    applied.


"... if that doesn't reduce the timeline beyond the root timeline."

maarten-ic added 6 commits July 15, 2026 13:13

Additional dunder methods for Ports and Timeline

3d1f169

Fix potential typing issue

40d6bef

Signature accepts a Sequence, but the logic previously only worked when providing a list. Other sequences silently didn't work.

Extend Timeline API

56ebfb2

Add `parent` property and `relative_to` (inspired by pathlib.Path)

Add timeline checker and testcases

a093eca

Update public API in v0.2/__init__.py

7a1a4e1

Rename test module

016a815

maarten-ic added this to 2026 MUSCLE3 improvements Jul 16, 2026

github-project-automation Bot moved this to Todo in 2026 MUSCLE3 improvements Jul 16, 2026

maarten-ic moved this from Todo to In Progress in 2026 MUSCLE3 improvements Jul 16, 2026

maarten-ic marked this pull request as ready for review July 17, 2026 10:15

maarten-ic requested review from IrisvdWerf and LourensVeen July 17, 2026 10:15

maarten-ic mentioned this pull request Jul 17, 2026

Add timeline determination and validation logic #27

Closed

3 tasks

maarten-ic added 3 commits July 17, 2026 13:41

Fix docstring link

f98cc6b

Merge remote-tracking branch 'origin/develop' into feature/resolve-ti…

c354722

…melines

Update timeline annotation in test

78280ff

LourensVeen requested changes Jul 20, 2026

View reviewed changes

Comment thread ymmsl/v0_2/tests/ymmsl1/timelines.ymmsl

Comment thread ymmsl/v0_2/tests/test_timeline_resolver.py

Comment thread ymmsl/v0_2/tests/ymmsl1/timelines.ymmsl

Comment thread ymmsl/v0_2/timeline_resolver.py Outdated

Comment thread ymmsl/v0_2/timeline_resolver.py Outdated

Improve exceptions, docstrings and add testcase

7f7add8

Based on PR feedback

maarten-ic requested a review from LourensVeen July 29, 2026 13:29

LourensVeen approved these changes Jul 29, 2026

View reviewed changes



		class CyclicDependency(RuntimeError):
		class ResolveTimelineException(RuntimeError):

Uh oh!

Conversation

maarten-ic commented Jul 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

maarten-ic commented Jul 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

maarten-ic commented Jul 17, 2026

Uh oh!

LourensVeen commented Jul 20, 2026

Uh oh!

LourensVeen left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

maarten-ic commented Jul 21, 2026

Uh oh!

LourensVeen commented Jul 29, 2026

Uh oh!

LourensVeen left a comment

Choose a reason for hiding this comment

Uh oh!

LourensVeen Jul 29, 2026

Choose a reason for hiding this comment

Uh oh!

LourensVeen Jul 29, 2026

Choose a reason for hiding this comment

Uh oh!

LourensVeen Jul 29, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

maarten-ic commented Jul 16, 2026 •

edited

Loading

maarten-ic commented Jul 16, 2026 •

edited

Loading