Re: How to show link types (was: Re: [Gzz] Re: Is xupdf reasonable?!)

[Asko Soukka]

Tue, 28 Jan 2003, Tuomas Lukka wrote:
> > >>>>>>http://www.workpractice.com/trigg/thesis-chap4.html
> > >>What exactly is the problem? I mean if we have link types (which we
> > >>don't have yet) and direction (which we already have), what else is there?
> > >
> > >We have direction (let's say left or right), but I thought we don't want
> > >to fix their meanings. E.g. I should be able to create "generalization"
> > >link either to the left or right from the curren cell, and the after that
> > >the link should be interpreted as "spesification" when going "backwards"
> > >to my creation direction.
> > >
> > >If we fix left->right = specification and right->left = generalization,
> > >then it's not problem, but AFAIK we don't want to do it.
> >
> > We could of course have two link types, for one of which left->right =
> > specification and for the other right->left = specification.
>
> ARGH! NO!
>
> People, this discussion is going towards systems SO complicated
> no-one would like to use them!

Possibility to write single labels on connections (only by pointing,
clicking and writing) is probably simple enough? Though, links of the same
type could be detected only by comparing strings.

Enough for me, but as I tried to show, IMHO, in its simplest version,
those "generalization/spesification" like links are problem.

Where did this start... Matti quoestioned, how a new user could know which
one of the two connected papers refers the other and which is the referred
one. (as I understood: without zooming the both first and making decisions
from the content)

Benja wrote:
> I do not believe that putting more conceptual weight into the system
> (e.g. by introducing a 'semantic direction' on top of the 'view
> direction') is easier. Especially since all these are offers-- you don't
> have to learn about all possible link types at ANY point, any more than
> you have to learn about all possible dimensions (you can't, even).

How much would we lose if we don't specify fixed link types, but the
user could only label the connections (maybe even put a label
to the both ends or something more general)?

How would this look in the structure? Will we create cells for
connections? (might confuse general views a lot)

-- 
Asko Soukka     | Taitoniekantie 9 A 603 | address@hidden
+358-40-8235947 | FIN-40740 JYVÄSKYLÄ    | http://www.iki.fi/asko.soukka/