Re: [ba-ohs-talk] Concept: Typed Versioning
So now two mechanisms need development. One to represent the dynamic linking
and separate meta logic for determining the syntax/semantic update behavior.
Seems that the dynamic linking mechanism should be generic, while provisions
for user control to declare domain specific update behavior would drive the
generic linking.
Sandy (01)
> From: Eric Armstrong <eric.armstrong@sun.com>
> Reply-To: ba-ohs-talk@bootstrap.org
> Date: Thu, 21 Mar 2002 12:43:02 -0800
> To: ba-ohs-talk@bootstrap.org
> Subject: [ba-ohs-talk] Concept: Typed Versioning
> Resent-From: <klausner@cubicon.com>
> Resent-To: s.klausner@attbi.com
> Resent-Date: Thu, 21 Mar 102 15:42:27 EDT
>
> Eugene Eric Kim wrote:
>
>> On Wed, 20 Mar 2002, Sandy Klausner wrote:
>>
>>> Could you point me to the portion of your Graph-based Data Model
>>> architecture that addresses document change; what happens to all the links
>>> that go in and out? Can you edit a document but preserve its links? What
>>> happens when you follow a link to a paragraph that has been erased?
>>
>> I didn't address this explicitly in the white paper, but I did address it
>> in a post I made to ohs-dev last April:
>>
>> http://www.bootstrap.org/lists/ohs-dev/0600.html
>
> Ah. Some beautiful stuff there.
>
> It strikes me that in the two scenarios mentioned in that post, the issue
> in each case is to distinguish between TYPED VERSIONS.
>
> In other words, the arc going from version1 to version 2 also requires
> type information, where the type is basically:
> * Syntax change
> * Semantic change
>
> If the meaning is the same but the wording is different,
> it's a syntax change. Otherwise, it's a semantic change.
>
> It seems to me that the automatically determining the nature of
> a change would be well within the capacity of present-day
> heuristic systems. It would CLEARLY be in the realm of
> language-parsing common-sense reasoning systems that
> can create a semantic model of a node. If the semantic model
> is unchanged, then it was a syntax change.
>
> Cases like:
> Its blue --> It's blue. (syntax)
> It's blue --> It's not blue. (semantics)
>
> Could be automatically discerned. A really smart system might
> even be able to tell when it didn't know, and ask for help in
> classifying a change.
>
> At worst, in the absence of such automated behavior, the user
> would have to manually specify change types. That would be
> too horrible to contemplate in production use, but it might be
> a way to get off the ground and see how "typed versioning"
> works.
>
> (02)