34
terleaved, a support model that works equally well for both kinds of work signifi-
cantly simplifies the problem.
3. It should provide a complete infrastructure for supporting single and multiple user
undo and redo. Authors need to be able to change their minds, and may need to
undo other authors’ changes in order to make their own. Not only should a system
support such activity, but it should also be able to track such changes, and possibly
revert them later. In the social give-and-take of editing, decisions should be tracked
and be capable of being revisited.
4. It should support flexible version management for version graphs with branching and
full merging (multiple ancestry).When changes are being made, significant states of
the document are frequently extremely important, and must be tracked. While ac-
commodating divergent states may be useful, most authorship tasks require the pro-
duction of a final version, so that integration of separate versions should be as easy
as is technically feasible.
5. It should provide reference mechanisms that allow the easy and consistent mainte-
nance of hypertext links, even across versions in a multi-user editing context.With the
WWW, hypertext writing is obviously moving from a special application into the
mainstream of document authoring; less obviously, linking features are built into
most current word processors. While inclusion of links in documents is now relatively
straightforward, maintaining links during a collaborative editing process is not. The
use of fine-grained linking as part of the collaboration process itself can also be very
important. This is already a part of current practice in the form of annotation fea-
tures.
6. The mechanism should, if possible, be extendable to cover a variety of general data
structures.Many documents are already multi-media documents (at least to the ex-
terleaved, a support model that works equally well for both kinds of work signifi-
cantly simplifies the problem.
3. It should provide a complete infrastructure for supporting single and multiple user
undo and redo. Authors need to be able to change their minds, and may need to
undo other authors’ changes in order to make their own. Not only should a system
support such activity, but it should also be able to track such changes, and possibly
revert them later. In the social give-and-take of editing, decisions should be tracked
and be capable of being revisited.
4. It should support flexible version management for version graphs with branching and
full merging (multiple ancestry).When changes are being made, significant states of
the document are frequently extremely important, and must be tracked. While ac-
commodating divergent states may be useful, most authorship tasks require the pro-
duction of a final version, so that integration of separate versions should be as easy
as is technically feasible.
5. It should provide reference mechanisms that allow the easy and consistent mainte-
nance of hypertext links, even across versions in a multi-user editing context.With the
WWW, hypertext writing is obviously moving from a special application into the
mainstream of document authoring; less obviously, linking features are built into
most current word processors. While inclusion of links in documents is now relatively
straightforward, maintaining links during a collaborative editing process is not. The
use of fine-grained linking as part of the collaboration process itself can also be very
important. This is already a part of current practice in the form of annotation fea-
tures.
6. The mechanism should, if possible, be extendable to cover a variety of general data
structures.Many documents are already multi-media documents (at least to the ex-
