Alb999 rail alignment value constraints #37

aothms · 2023-01-22T14:06:11Z

Rule implementation

An IfcAlignmentHorizontal shall only nest a list of IfcAlignmentHorizontalSegment (through IfcAlignmentSegment) that has PredefinedType set to LINE, CIRCULARARC or CLOTHOID.

NB The validation service validates requirements that are universally true and not organization or context-specific. For that purpose:

This rule still needs to be updated to be universally true, with
- context-specific requirements encoded in IDS (this is possible because the non-rooted entity IfcAlignmentHorizontal can be selected with an attribute constraint on PredefinedType)
- the rule remainig universal but individual stakeholders "forking" the repository to make organization-specific adaptations
(or) the rule will remain just an example and will not be merged to the main branch (but can still be invoked through the "sandbox" functionality)

Considering the following example instantiation diagram

the feature has been annotated with the context (name of instance) being operated on in square brackets ([ ... ])

@implementer-agreement
@ALB
Feature: Rail Alignment value constraints

  Background: Rail alignment
  
[AL1, AL2]           Given An IfcAlignment
[AL1]                And A relationship IfcRelReferencedInSpatialStructure to IfcAlignment from IfcRail
    
  Scenario: Agreement on allowed values for horizontal segment types
    
[AL1_H]              Given A relationship IfcRelNests from IfcAlignment to IfcAlignmentHorizontal and following that
[AL1_H_S1, AL1_H_S2] Given A relationship IfcRelNests from IfcAlignmentHorizontal to IfcAlignmentSegment and following that
[..., ...]           Given Its attribute DesignParameters
[LINE, HELMERTCURVE] Given Its attribute PredefinedType
    
[raises error        Then the value must be "LINE" or "CIRCULARARC" or "CLOTHOID"
on HELMERTCURVE]

The current validation report for a population model like this is according to:

Rail Alignment value constraints/Agreement on allowed values for horizontal segment types.v1
• the value must be "LINE" or "CIRCULARARC" or "CLOTHOID"
The value 'HELMERTCURVE' on #40=IfcAlignmentHorizontalSegment('17KlfolOr6ReLWMUjUKk$j',$,$,$,$,$,$,$,.HELMERTCURVE.) is not one of 'LINE', 'CIRCULARARC', 'CLOTHOID'

In principal the entire history of "stack frames" is available when building the report so we can recover the full path from alignment to segment type.

Background: Rail alignment

This rule uses the background statement, which is functionally equivalent to having the same given statements prepended to each of the scenarios, but offers additional advantages later in categorizing such rules based on their background. We might be referring to this background selection as an "activation mechanism", i.e a graph pattern (in this case) that triggers validation constraints to become applicable.

to/from

A relationship IfcRelReferencedInSpatialStructure to IfcAlignment from IfcRail
                                                  ^^

to refers to the Relating(Object/...) side of the relationship node

to and from can be positioned arbitrarily but the convention is that the first end of the relationship should correspond to the previously established context.

corresponding nodes to build a graph

Given An IfcAlignment
         ^^^^^^^^^^^^
  And A relationship IfcRelReferencedInSpatialStructure to IfcAlignment from IfcRail
                                                           ^^^^^^^^^^^^

The first part of the relationship definition should correspond to the previously established context

traversal

Given An IfcAlignment
         ^^^^^^^^^^^^
  And A relationship IfcRelReferencedInSpatialStructure to IfcAlignment from IfcRail
                                                           ^^^^^^^^^^^^
...

Given A relationship IfcRelNests from IfcAlignment to IfcAlignmentHorizontal and following that
                                      ^^^^^^^^^^^^

In the first case the context stays on alignment, so it gets repeated in the third statement.

Given A relationship IfcRelNests from IfcAlignment to IfcAlignmentHorizontal and following that
                                                                             ^^^^^^^^^^^^^^^^^^
Given A relationship IfcRelNests from IfcAlignmentHorizontal to IfcAlignmentSegment and following that

However with the and following that clause we cause the context to switch to IfcAlignmentHorizontal (in this case).

…n of gherkin rule

typo Co-authored-by: Stefan Jaud <59165496+pjanck@users.noreply.github.com>

…fier' into rail_alignment

…lue_constraints

aothms · 2023-01-22T14:13:04Z

Sandbox ok ca8bf07

aothms · 2023-01-22T14:24:29Z

Sandbox ok c3b4a2c

aothms · 2023-01-22T14:28:29Z

Sandbox ok e2d5cfa

pjanck

Initial comments for Gherkin.

pjanck · 2023-01-22T19:47:00Z

features/ALB999_rail_alignment_value_constraints.feature

+    Given A relationship IfcRelNests from IfcAlignment to IfcAlignmentHorizontal and following that
+    Given A relationship IfcRelNests from IfcAlignmentHorizontal to IfcAlignmentSegment and following that
+    Given Its attribute DesignParameters
+    Given Its attribute PredefinedType


Did you meant And in the last three lines? Or are all four conditions to be interpreted independently of one another?

There is no functional difference between that. and is just a shortcut for the previously used gherkin keyword. But I agree it should be used consistently (maybe it's actually clearer to never use and).

pjanck · 2023-01-22T19:51:08Z

features/OJP001_Relative-placement-for-elements-aggregated-to-another-element.feature

+      And An IfcElement
+      And A relationship IfcRelAggregates from IfcElement to IfcElement
+      Then The relative placement of that IfcElement must be provided by an IfcLocalPlacement entity
+      And The PlacementRelTo attribute must point to the IfcLocalPlacement of the container element established with IfcRelAggregates relationship


What if the container is placed by IfcGridPlacement? As the sentence reads right now, it is not allowed for IfcRelAggregates.RelatingObject to ever be placed by anything else as IfcLocalPlacement.

pjanck · 2023-01-22T19:53:35Z

features/ALB999_rail_alignment_value_constraints.feature

+    Given A relationship IfcRelNests from IfcAlignment to IfcAlignmentHorizontal and following that
+    Given A relationship IfcRelNests from IfcAlignmentHorizontal to IfcAlignmentSegment and following that


What is and following that?

Good question, I've updated the PR description. You can find the answer under "traversal". In short it means that, in the first case without and following that the a relationship ... statement checks whether such a relationship exists, but keeps the context on the previously active node (alignment). In the second case we actually descend according to the nesting structure, so that the attribute DesignParameters is read from the IfcAlignmentSegment.

That makes a lot of sense, nice to see that this approach is useful for more rules (also referring to your explanation in the top of this PR). For the 0.6 sprint we are encountering a rule (briefly discussed in a previous meeting) for which we will need attributes for both the previously activate node and the following elements. That seems to be possible this way. Or maybe we will need to make it more explicit.

pjanck · 2023-01-22T19:58:17Z

features/OJP001_Relative-placement-for-elements-aggregated-to-another-element.feature

+      And A relationship IfcRelAggregates from IfcElement to IfcElement
+      Then The relative placement of that IfcElement must be provided by an IfcLocalPlacement entity


Which that does this sentence refer to? In the sentence above, there are two IfcElements ...

I'm not going to reply here, this feature shows up here, because my PR was based on two un-merged PRs. See #27.

Ghesselink · 2023-01-23T18:21:39Z

features/ALB999_rail_alignment_value_constraints.feature

+    Given A relationship IfcRelNests from IfcAlignment to IfcAlignmentHorizontal and following that
+    Given A relationship IfcRelNests from IfcAlignmentHorizontal to IfcAlignmentSegment and following that


That makes a lot of sense, nice to see that this approach is useful for more rules (also referring to your explanation in the top of this PR). For the 0.6 sprint we are encountering a rule (briefly discussed in a previous meeting) for which we will need attributes for both the previously activate node and the following elements. That seems to be possible this way. Or maybe we will need to make it more explicit.

Ghesselink · 2023-01-23T18:27:37Z

features/steps/steps.py

+
+        if constraint in ('identical', 'unique'):
+
+            for i, values in enumerate(instances):
+                if not values:
+                    continue
+
+                msg = value_error_msg(identical_or_unique=constraint, attribute=getattr(context, 'attribute', 'None'))


Is it okay for you to convert this draft PR to an actual PR (without merging it)? I would like to test where the code is failing for the open PRs, but cannot git checkout in a branch for this PR.
Or is there no branch for this PR?

Ghesselink · 2023-01-23T18:31:54Z

features/steps/steps.py

+register_type(from_to=TypeBuilder.make_enum({"from": 0, "to": 1 }))
+register_type(maybe_and_following_that=TypeBuilder.make_enum({"": 0, "and following that": 1 }))
+
+


It computes a regular expression pattern for the given enumeration of words/strings and stores them in X.pattern attribute.

That is very useful

Ghesselink · 2023-01-23T18:59:10Z

features/steps/steps.py

+@given('A relationship {relationship} {dir1:from_to} {entity} {dir2:from_to} {other_entity}')
+@given('A relationship {relationship} {dir1:from_to} {entity} {dir2:from_to} {other_entity} {tail:maybe_and_following_that}')
+def step_impl(context, relationship, dir1, entity, dir2, other_entity, tail=0):
+    assert dir1 != dir2


Nice implementation :-) Is it an idea to also assert it is in the right direction? I would think if someone working on this would break the convention you were describing, it would return an empty list context.instances, resulting in a false positive. I can imagine it would help with debugging for someone writing a rule, but perhaps the effort is greater than the benefit in this.

aothms · 2023-01-25T20:32:34Z

Sandbox ok 08ae8d6

aothms · 2023-01-27T10:19:25Z

Sandbox ok 45690e3

aothms · 2023-01-27T13:49:28Z

Sandbox ok 427fbf0

Ghesselink · 2023-01-28T01:57:40Z

features/environment.py

 def before_feature(context, feature):
+    # @tfk we have this strange issue between stack frames blending over
+    # between features so we need to preserve only the bottom two stack
+    # frames when beginning a new feature.
+    context._stack = context._stack[-2:]


Aah, that explains it .. I tried to add context._push() with at the 'Its values' statement to retrieve it later and run into exactly this (another feature somehow not working anymore ..).

JTurbakiewicz and others added 17 commits December 12, 2022 00:13

Initial commit of OJP001 rule

887e60e

initial commit

bc748b1

incorporate discussion PR for both generating files and implementatio…

cca7b3a

…n of gherkin rule

gherkin statements with stack frames

d1c1dfb

add map_state, return values from stack with indices

8a33377

Update features/GEM003_Unique-representation-identifier.feature

50e65d8

typo Co-authored-by: Stefan Jaud <59165496+pjanck@users.noreply.github.com>

Changes and simplifications

f1501a7

shorten gherkin statement

7df2ab7

align with grf001

a19d839

remove random string

3044bad

Merge branch 'main' into gem003_unique-representation-identifier

5272bcd

MR comments, added lookup table, changed the wording

7563038

Changed relationships from dict to csv + MR comments

287a576

Parametrize rel direction

d84ff3a

Merge remote-tracking branch 'bsi/gem003_unique-representation-identi…

5501255

…fier' into rail_alignment

Generalize relationship stmt, value constraint

58c942b

Merge remote-tracking branch 'bsi/main' into ALB999_rail_alignment_va…

ca8bf07

…lue_constraints

aothms requested review from johltn, Ghesselink and JTurbakiewicz January 22, 2023 14:06

aothms marked this pull request as draft January 22, 2023 14:07

Fix path syntax

c3b4a2c

Add feature

e2d5cfa

pjanck suggested changes Jan 22, 2023

View reviewed changes

Ghesselink reviewed Jan 23, 2023

View reviewed changes

aothms added 2 commits January 25, 2023 21:26

Test files

0b67572

ALB999 IfcAlignment start station

08ae8d6

aothms added 2 commits January 27, 2023 11:12

Clear stack frames in between features

a488fe3

ALB002 account for None

45690e3

ALB002 Complete schema version applicabilities

427fbf0

Ghesselink reviewed Jan 28, 2023

View reviewed changes

Ghesselink mentioned this pull request Feb 1, 2023

GEM005 representation in decomposed elements #40

Closed

aothms mentioned this pull request Sep 21, 2023

SYS001 / RI-309 cable signal flow #84

Merged

Ghesselink mentioned this pull request Apr 14, 2024

simplify retrieval of activation instances #166

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Alb999 rail alignment value constraints #37

Alb999 rail alignment value constraints #37

aothms commented Jan 22, 2023 •

edited

aothms commented Jan 22, 2023

aothms commented Jan 22, 2023

aothms commented Jan 22, 2023

pjanck left a comment

pjanck Jan 22, 2023

aothms Jan 23, 2023

pjanck Jan 22, 2023

pjanck Jan 22, 2023

aothms Jan 23, 2023 •

edited

Ghesselink Jan 23, 2023 •

edited

pjanck Jan 22, 2023

aothms Jan 23, 2023

Ghesselink Jan 23, 2023 •

edited

Ghesselink Jan 23, 2023

Ghesselink Jan 23, 2023 •

edited

Ghesselink Jan 23, 2023

aothms commented Jan 25, 2023

aothms commented Jan 27, 2023

aothms commented Jan 27, 2023

Ghesselink Jan 28, 2023

		Given A relationship IfcRelNests from IfcAlignment to IfcAlignmentHorizontal and following that
		Given A relationship IfcRelNests from IfcAlignmentHorizontal to IfcAlignmentSegment and following that

		And A relationship IfcRelAggregates from IfcElement to IfcElement
		Then The relative placement of that IfcElement must be provided by an IfcLocalPlacement entity

		register_type(from_to=TypeBuilder.make_enum({"from": 0, "to": 1 }))
		register_type(maybe_and_following_that=TypeBuilder.make_enum({"": 0, "and following that": 1 }))

Alb999 rail alignment value constraints #37

Are you sure you want to change the base?

Alb999 rail alignment value constraints #37

Conversation

aothms commented Jan 22, 2023 • edited

Rule implementation

Background: Rail alignment

to/from

corresponding nodes to build a graph

traversal

aothms commented Jan 22, 2023

aothms commented Jan 22, 2023

aothms commented Jan 22, 2023

pjanck left a 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

aothms Jan 23, 2023 • edited

Choose a reason for hiding this comment

Ghesselink Jan 23, 2023 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Ghesselink Jan 23, 2023 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Ghesselink Jan 23, 2023 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

aothms commented Jan 25, 2023

aothms commented Jan 27, 2023

aothms commented Jan 27, 2023

Choose a reason for hiding this comment

aothms commented Jan 22, 2023 •

edited

aothms Jan 23, 2023 •

edited

Ghesselink Jan 23, 2023 •

edited

Ghesselink Jan 23, 2023 •

edited

Ghesselink Jan 23, 2023 •

edited