home *** CD-ROM | disk | FTP | other *** search
Text File | 1989-06-26 | 3.8 KB | 70 lines | [TEXT/GEOL] |
- Item 1260066 8-June-89 11:41
-
- From: ALGER Alger, Jeff
-
- To: MACAPP.TECH$ MACAPP Tech
-
- Sub: Re: TDocument (again)
-
- Carl et Al..
-
- Some additional thoughts on the TDocument issue. I agree totally with the need
- for some visual object for the user to manipulate; checking options in a menu
- is too ... well, it's just not a Mac approach. Where I part ways is with the
- concept of a TDocument as the be all and end all of the interface and the
- internals, perhaps even the name "document" itself. To me, a "document" should
- be something I can print. That applies to a word processing document or a
- drawing. It does not apply to more complicated aggregations of data such as
- databases. There is no visual, hardcopy analogy to a database; only to
- portions of that database as retrieved by a query and arranged by one or more
- programs.
-
- The other problem I have with the concept of a TDocument as it exists today is
- that it encourages a close tie between the organization of data as perceived by
- the user (i.e., windows) with the underlying sources and sinks of data
- (DoRead/DoWrite methods). Several of the links suggest refinements to
- TDocument to allow, say, one TDocument to "control" several TFile or whatever
- instances. Yet, this still clings to a hierarchy in which the user's perceived
- organization is a partition of the underlying data, a concept which is in many
- cases simply not valid.
-
- Consider a relational DBMS which allows relations (files) to be virtual; i.e.,
- logically defined and derived at run time rather than physically stored. One
- such logical view (view in DBMS parlance, not MacApp) of the data may combine a
- local database with one or more remote ones, or several local ones. The only
- thing one can describe as "the" document is the entire interface to the
- database(s), which to a user is indistinguishable from the application itself.
- I claim that this is not at all useful in presenting the data to the user, who
- should be as oblivious as possible to the underlying processing and structures.
- Indeed, the application program itself may be unaware of how these views are
- constructed, particularly with gateway products like CL/1!
-
- The most natural "document" for the user may be a report or a data entry
- window, both of which can be printed, while the most natural "document" for the
- program, if there is one, is the query or the relation. Neither necessarily
- corresponds to the underlying structure of the database(s). To insist on
- mapping these three into some sort of TDocument-controlled hierarchy is
- difficult and without merit. It also flies in the face of the experience of
- the database community, which is that databases should be designed and
- maintained independently of the applications which use them. "Documents" in
- the Mac interface are creatures of the application; the data may have an
- independent life and may, in fact, appear differently to different users and
- applications, both in structure and content.
-
- In short, let TDocument be a creature of the user interface and develop richer
- - and more appropriate - classes to deal with internal organization. I have
- used databases for my soapbox throughout my links, but I feel that the
- principle is valid in non-database environments as well. Any time the physical
- organization of data does not correspond to the printed - and, by implication,
- visual - organization, or when multiple applications share the same data, the
- current TDocument usage is trouble. This happens often enough that it should
- be split in two.
-
- As to specific suggestions for what that underlying structure should be, I will
- need a couple of days to organize my thoughts and will then send a follow-up
- link. In the meantime, what does everyone else think?
-
- Jeff Alger
- Peat Marwick Main & Co.
-
-