45
actions.
1
This information, while it may be strictly redundant in the sense of version-
completeness, can prove critical in preserving users’ editing intentions when histories are
modified. Operations like merge and selective undo depend critically on the actual opera-
tions performed, and are one of the reasons for adopting change-complete operation models.
In a version-complete system, the result of merging two other states is not in itself spe-
cial: any change set is acceptable that correctly produces an acceptable result: representing
the saved states of the shared structure while meeting the complexity and space require-
ments of the application. In a typical operation-complete system, such a result state would
be the result of some kind of merge operation performed on the operation histories that cre-
ated the two relevant versions, and it would reflect the combined effects of the primitive
operations in those histories.
2.4 Limits of the taxonomy
In the following, I summarize some key aspects of change interaction by enumerating
types of interaction and conflict resolution. The global context of operation types, potential
inter-operation interactions, and potential conflicts provides a conceptual framework for un-
derstanding the conflict situations in Chapter 3,and the design decisions in the Palimpsest
model presented in Chapter 4.
In this taxonomy I consider properties of operations that are independent of the policy
issues which arise in the implementation of facilities like merge and undo. These facilities
create serious issues of change conflict that we will discuss in Chapter 3,in the context of a
1
These are distinct, in that most user’s mental models of editing include operations like “move,”
but their actual actions in most editors are a combination of “cut” and “paste” actions. Change repre-
sentations oriented to human needs are more important than application-oriented ones, where the
distinction is possible.
actions.
1
This information, while it may be strictly redundant in the sense of version-
completeness, can prove critical in preserving users’ editing intentions when histories are
modified. Operations like merge and selective undo depend critically on the actual opera-
tions performed, and are one of the reasons for adopting change-complete operation models.
In a version-complete system, the result of merging two other states is not in itself spe-
cial: any change set is acceptable that correctly produces an acceptable result: representing
the saved states of the shared structure while meeting the complexity and space require-
ments of the application. In a typical operation-complete system, such a result state would
be the result of some kind of merge operation performed on the operation histories that cre-
ated the two relevant versions, and it would reflect the combined effects of the primitive
operations in those histories.
2.4 Limits of the taxonomy
In the following, I summarize some key aspects of change interaction by enumerating
types of interaction and conflict resolution. The global context of operation types, potential
inter-operation interactions, and potential conflicts provides a conceptual framework for un-
derstanding the conflict situations in Chapter 3,and the design decisions in the Palimpsest
model presented in Chapter 4.
In this taxonomy I consider properties of operations that are independent of the policy
issues which arise in the implementation of facilities like merge and undo. These facilities
create serious issues of change conflict that we will discuss in Chapter 3,in the context of a
1
These are distinct, in that most user’s mental models of editing include operations like “move,”
but their actual actions in most editors are a combination of “cut” and “paste” actions. Change repre-
sentations oriented to human needs are more important than application-oriented ones, where the
distinction is possible.