[library] oclstdlib.uml validation failures

[eclipse-ocl-bot]

| --- | --- | | Bugzilla Link | 287010 | | Status | CLOSED WONTFIX | | Importance | P3 normal | | Reported | Aug 19, 2009 01:43 EDT | | Modified | May 25, 2015 17:21 EDT | | Version | 1.3.0 | | Depends on | 321461 | | Blocks | 318368 | | Reporter | Ed Willink |

Description

[Using the Galileo Modeling release and (the same) OCL from CVS.]

Invoking Validate within the UML Model Editor for the oclstdlib package in oclstdlib.uml gives 2 validation errors. (Invoke again if errors don't show; there is something intermittent.)

The 'checkExclusiveSignature' invariant is violated on 'Message Type OclMessage'

The required feature 'referredType' of 'Type Type' must be set

Clearly the checkExclusiveSignature diagnostic could be improved to give more information, since everything looks good.

[eclipse-ocl-bot]

By Ed Willink on Aug 19, 2009 02:21

Same problems in oclstdlib.ecore (plus ill-formed name issues raised in Bug 287014).

[eclipse-ocl-bot]

By Ed Willink on Aug 01, 2010 10:45

For Helios, there 435 errors on ocl.uml; mostly due to missing base_XXX features.

For Helios, there 188 on oclstdlib.uml; mostly due to inconmsistent opposites (on operations!).

The 'intermittency' is probably because of inadvertant use of the rival 'Validate' unhelpfully contributed by WTP in Bug 319945, Bug 197617.

[eclipse-ocl-bot]

By Kenn Hussey on Aug 19, 2010 15:00

(In reply to comment #2)

For Helios, there 435 errors on ocl.uml; mostly due to missing base_XXX features.

I would recommend setting the lower bound for these properties (and in turn, their Ecore counterparts) to 0.

[eclipse-ocl-bot]

By Ed Willink on Nov 08, 2011 04:26

The problems look to be the same in Juno M3.

(In reply to comment #3)

(In reply to comment #2)

For Helios, there 435 errors on ocl.uml; mostly due to missing base_XXX features.

I would recommend setting the lower bound for these properties (and in turn, their Ecore counterparts) to 0.

I'm not clear how this can be done since where a lower bound is possible on a EGenericType, it must be a type, and setting an arbitrary tupe makes no difference anyway.

This causes no problem to users and is now legacy code for developers. I'm inclined to WONTFIX.

[eclipse-ocl-bot]

By Ed Willink on May 27, 2014 10:50

(In reply to Ed Willink from comment #4)

This causes no problem to users and is now legacy code for developers. I'm inclined to WONTFIX.

Only 70 errors in Luna.

[eclipse-ocl-bot]

By Ed Willink on May 25, 2015 17:21

CLOSED after more than a year in the RESOLVED state.