Align with UML Specification: Literals #3749

pbrown12303 · 2025-02-03T15:07:49Z

This change brings the core structure of Gaphor UML into alignment with the UML 2.5 specification. It does not implement all of the UML specification. There are two significant areas of modification:

The <<simpleAttribute>> stereotype was removed from ValueSpecification. This required implementing the Literal subclasses of ValueSpecification, with the exception of LiteralReal. Attributes that previously referenced ValueSpecifications as simple values have been converted to references to ValueSpecifications. This required updating all code that accessed these references. Recipes have been added to facilitate accessing and updating these references. The scope of this change was fairly broad.
All relationships have been aligned with the UML specification. This particularly impacted parent-child relationships. An update to the property subset logic was required. It should be noted that in a couple of places the UML specification has two different relationships between the same two classes that use the same role name. StructuredClassifier <--> Property and Property <--> Association are two examples. In both cases, one of the relationships had to be renamed as Gaphor treats the role names as class property names, which led to ambiguity.

The storage.py file has been updated to make these changes backwards compatible.

PR Checklist

Please check if your PR fulfills the following requirements:

I have read, and I am following the Contributor guide
I have read, and I understand the GNOME Code of Conduct

PR Type

What kind of change does this PR introduce?

Bug fix
Feature
Chore (refactoring, formatting, local variables, other cleanup)
Documentation content changes

What is the current behavior?

Issue Number: #3720, #3517,

What is the new behavior?

Does this PR introduce a breaking change?

Yes
No

Other information

There are some issues: * I cannot set default values for attributes in Gaphor

amolenaar · 2025-02-05T16:14:54Z

I merge the main branch into this PR. In the process it looks like I introduced an issue: I cannot set a default value on an attribute anymore.

pbrown12303 · 2025-02-05T16:21:16Z

Thanks @amolenaar - I couldn't figure out the git incantation to do this merge on my machine. I'll take a look and fix the problem.

amolenaar · 2025-02-05T18:08:56Z

In also want to make another attempt.

It's a bit of a pity though that all changes are in one commit. If it were multiple smaller commits, merge conflicts are easier to resolve.

amolenaar · 2025-02-05T21:21:41Z

gaphor/C4Model/c4model.py

@@ -4,6 +4,8 @@

 from __future__ import annotations

+from decimal import Decimal as UnlimitedNatural


Why use Decimal as UnlimitedNatural?

A natural number is whole number right? If so, we can use int: integers in Python grow and do not overflow, similar to BigInteger in Java.

The int type can't handle "*". I used Decimal with conversions in and out. Can't see any other way.

If I could create a type as something like "int | math.inf", that would do the trick, but math.inf is not a type. The only types in Python that can handle infinite values are Decimal and Float. I chose Decimal. Of course, this requires constraining the Decimal input values to int or math.inf, and constraining the Decimal values to str(int(value)) or "*" when converting back to a string. I don't really see any alternatives - do you?

We can use a union type int | typing.Literal["*"] instead.

Interesting. Let me try that.

gaphor/codegen/coder.py

gaphor/core/modeling/coremodel.py

amolenaar · 2025-02-05T21:54:00Z

gaphor/core/modeling/properties.py

+        if self.type is decimal.Decimal and isinstance(value, str):
+            if value == "*" or value == "inf" or value == "Infinity":
+                value = decimal.Decimal(math.inf)
+            else:
+                value = decimal.Decimal(int(value))


Not sure why we need decimal here. We deal with natural (whole, positive) numbers.

See earlier comment abut UnlimitedNatural. The issue is handling infinite values.

pbrown12303 · 2025-02-06T20:47:39Z

This is bizarre. The test that failed on windows (in the teardown) passes when I run it in the debugger in my development environment (which is where and how I usually run the tests), but it fails when I run it without the debugger! I'm looking at it now.

amolenaar · 2025-02-12T21:56:30Z

gaphor/core/modeling/properties.py

+        if opposite is not None:
+            self.original.opposite = opposite


We should probably not change the original opposite relation when a relation is redefined.

I assume the opposite relation should wither point to the original opposite end, or a redefined one.

I removed this code, and instead make the relation Property.class_ <-> Class.ownedAttribute redefined Property.structuredClassifier <-> StructuredClassifier.ownedAttribute.

Property.class_ did not redefine Property.structuredClassifier, but was a subset, whereas Class.ownedAttribute redefined StructuredClassifier.ownedAttribute. Both ends should redefine their "super parts" for this to work.

See 3dc4747.

amolenaar · 2025-02-13T14:09:42Z

gaphor/UML/recipes.py

+    assoc.ownedEnd = end_a
+    assoc.ownedEnd = end_b


Why has this changed to ownedEnd? Shouldn't an association end be at least a memberEnd of an association? Otherwise Property.association is not set, which should be set always if a property belongs to an association, if I understand correctly.

Setting ownedEnd has no effect, since the change is undone by set_navigability.

Changed this: Association.ownedEnd is set by the set_navigability function. To my understanding, Property.association should referenjce the association.

See 3dc4747

Redefined properties should not overwrite opposite end of original Property.class_ pointer to Class.ownedAttribute, but Class.ownedAttribute pointed to Property.structuredClassifier. In case an association is birections, both ends should be redefined.

It's still used for expressions. This reverts commit 56619b6.

amolenaar

I think we can merge this.

The use of decimal.Decimal is internal, so we can factor that out in a later PR.

amolenaar · 2025-02-19T12:23:22Z

@pbrown12303 Thanks again for your help on making Gaphor the best modeling tool out there :)

pbrown12303 · 2025-02-19T17:41:48Z

@amolenaar You are quite welcome. I hope to add more. Starting my research now...

Align with UML Specification - all tests pass

b7dd610

github-actions bot added the python Pull requests that update Python code label Feb 3, 2025

pbrown12303 and others added 2 commits February 3, 2025 14:30

Updated association documentation

c9e3d35

Merge branch 'main' into uml-updates

cf052dd

There are some issues: * I cannot set default values for attributes in Gaphor

amolenaar requested changes Feb 5, 2025

View reviewed changes

pbrown12303 added 2 commits February 6, 2025 10:00

Fixed default value issues

26f9909

Further tweaks to values

f17dc0f

pbrown12303 and others added 3 commits February 6, 2025 15:56

Fixed bug in disconnect

27e2040

Add missing primitive to ValueSpecification mappings

ae60cb5

Fix diagram type inheritance for UML diagrams

ef2dd2b

amolenaar reviewed Feb 12, 2025

View reviewed changes

More consistent type checks: check for package first

92bb9a8

amolenaar reviewed Feb 13, 2025

View reviewed changes

amolenaar added 4 commits February 13, 2025 16:27

Remove SimpleAttribute stereotype from UML model

56619b6

Revert "Remove SimpleAttribute stereotype from UML model"

44a5749

It's still used for expressions. This reverts commit 56619b6.

Fix copy of valuespecifications for actions, classes and deployments

df6ca2d

amolenaar self-requested a review February 14, 2025 09:59

amolenaar approved these changes Feb 14, 2025

View reviewed changes

amolenaar merged commit d96c4fd into gaphor:main Feb 19, 2025
24 checks passed

pbrown12303 deleted the uml-updates branch February 19, 2025 17:42

amolenaar changed the title ~~Align with UML Specification - all tests pass~~ Align with UML Specification: Literals Apr 16, 2025

danyeaw added feature A new feature and removed python Pull requests that update Python code labels May 20, 2025

		@@ -4,6 +4,8 @@

		from __future__ import annotations

		from decimal import Decimal as UnlimitedNatural

Uh oh!

Align with UML Specification: Literals #3749

Align with UML Specification: Literals #3749

Uh oh!

Conversation

pbrown12303 commented Feb 3, 2025

PR Checklist

PR Type

What is the current behavior?

What is the new behavior?

Does this PR introduce a breaking change?

Other information

Uh oh!

amolenaar commented Feb 5, 2025

Uh oh!

pbrown12303 commented Feb 5, 2025

Uh oh!

amolenaar commented Feb 5, 2025

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!

pbrown12303 commented Feb 6, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

amolenaar Feb 13, 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!

amolenaar left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

amolenaar commented Feb 19, 2025

Uh oh!

pbrown12303 commented Feb 19, 2025

Uh oh!

Uh oh!

amolenaar Feb 13, 2025 •

edited

Loading