Inferred user entry of data

[mike1813]

Construction pattern HuirPr-vD-cC-iHu-DS+iD infers that a Human interactive user supplies Data to a Process where the Process receives the Data, but it is not entered by any other Human, created by any other Process , nor stored on any Host. The assumption is that as no other source for the Data exists, it must be provided by the Human user.

However, where the Process 'reads' the Data, the Process (which in principle includes the user interaction) can't proceed without access to this Data. In that situation there is some ambiguity. Should we assume:

• that the Human does provide the missing Data, as currently implemented, or
• that the Process including the user will be stuck?

The original plan was to make the latter assumption, but pattern HuirPr-vD-cC-iHu-DS+iD fails to check there is no 'reads' dependency between the Process and the Data.

[mike1813]

Made the necessary changes to HuirPr-vD-cC-iHu-DS+iD, but it turned out this pattern covers all Processes including Editors. The original plan was to assume that if the Process is an Editor (i.e., a process whose ONLY function is data entry or update), then the user does supply Data if it could come from nowhere else, even where there is an asserted 'reads' dependency.

The solution was to create a duplicate pattern in which the Process is restricted to Editor subclasses, in which the prohibited 'reads' relationship is not included.

[mike1813]

Addressed in branch 65, as this issue affects some system models used as regression test cases.