Hi,
I have discussed the previous week with Nicolas at FOSDEM and he has
convinced NSDataLink related classes (from OpenStep) partially
implemented in GNUstep currently (but not in Cocoa), could provide us a
simple and efficient way to add document orientation to Étoilé
applications… It would be a first step before thinking how to move to a
more evolved components architecture later.
yep :-)
for thoses that don't know NSDataLink, it let you link data instead of
just paste it; and you have different possibilities to handle the way
the data is updated (real time, etc.)
Basically that's what Steve Jobs is demo-ing in the recently diffused
NeXTSTEP 3.0 video.
- First step would be to add basic document orientation without
changing GNUstep applications UI interaction model.
Wouldn't it be taking care of by NSDocument ?
- Next step would be to tailor the UI in order to better fit the
document orientation paradigm.
what do you mean by that ? the "document icon" on the titlebar ?
- Last step would be the components architecture (with set of loadable
components to customise skeleton-like applications for advanced tasks
and your own workflow)
yes. That part is not handled by NSDatalink. Personally I think that
such a OLE-like technology it's not _that_ useful for having a
document-based UI -- NSDatalink or something similar will work
reasonably the same, yet is magnitude simpler to have.
But of course, an OLE-like architecture (even better hopefully ;-) also
has its place and would be nice to have. It would be interesting to
have a look to KParts too..
I have just taken a look at the new LinkBack
<http://www.linkbackproject.org/> project initiated by Nisus (with an
open source framework), it looks fairly similar to the first goal I
have just outlined over, but the framework is currently limited
compared to NSDataLink stuff, then it looks strange Nisus didn't choose
to reimplement NSDataLink & co on their side rather to create new API
from scratch.
Yes, I had a quick look on it too..
I guess the reason they didn't choose to implement NSDatalink is
because the Nisus guys probably didn't know about NSDatalink. And
perhaps they want something simpler (although they mention on their
page the scenario 2, live update...) ..
I need to read the documentation and the code more attentively however
I admit :-)
Anyway I'm quite sure we could implement LinkBack compatiblity on top
of NSDataLink & co.
good idea :-)
Actually, it could be interesting to send some mails to Nisus, tell
them about NSDatalink, the fact that it's currently going to be
implemented on gnustep, and see why they didn't choose that path.
Perhaps it's simply because they didn't know about it, and then we
could perhaps have some cooperation, or perhaps they have good reasons
that we could be interested in :-)
Cheers,
--
Nicolas Roard
"Any sufficiently advanced technology is indistinguishable from magic."
-Arthur C. Clarke
