home *** CD-ROM | disk | FTP | other *** search
Text File | 1990-03-09 | 1.7 KB | 47 lines | [TEXT/GEOL] |
- Item 3457257 7-March-90 18:44PST
-
- From: POWERUP.DEV Power Up Software,PRT
-
- To: MACAPP.TECH$ MacApp Technical
- ALGER Alger, Jeff,VCA
-
- Sub: Re- Re- Re- Model/View…
-
- Attn: MacApp.Tech$
- Attn: Alger, Jeff,VCA
- SentBy: James Plamondon
- Date 3/7/90
- Subject Re- Re- Re- Model/View…
- From James Plamondon
- To Alger, Jeff,VCA
- CC MacApp.Tech$
-
- Reply to: Re: Re: Re: Model/View…
- Jeff,
-
- OK, the TPublisher/TSubscriber mechanism sounds reasonsble. (I still like the
- Supporter and Dependent terminology better, but — what the heck.) I have a
- couple of questions left, though.
-
- I am afraid that I don't understand the advantage of having new classes stuck
- into the hierarchy over modifying the MacApp source directly. MacApp source
- isn't sacred. Rather than trying to wedge TSubscriber between TObject and
- TEventHandler, why not just add the necessary code to TEventHandler and punt
- the whole problem? Perhaps I'm missing something here.
-
- Likewise, the separation of TSubscriber and TPublisher into separate classes
- enforces what I cannot help but think is an unnecessary restriction: that a
- publisher cannot be a subscriber, and vice versa. By adding both the
- TPublisher and TSubscriber code directly to TObject, any object can then be
- either a publisher or a subscriber — or both. The only added overhead that I
- can see is in 1) bigger TObject method tables, and 2) the global dependency
- dictionary.
-
- Is there something specifically wrong or inadequate in this more general
- mechanism that would make it either impossible or inappropriate to implement?
-
- Looking forward to your response, I remain
-
- James Plamondon
-
-