Tomas Sanchez Romani wrote:
> What I'm after is not primarly a usable dtp-system , wysiwyg or
> "the-unix-way-with-lots-of-independent-comandline-programs";-)
Actually, the programs I use are pretty-much wysiwyg. LyX
is near wysiwyg, and Sketch certainly is. I do use several
independent programs, but some progress has been made towards
integrating them -- such as the Sketch script which calls
TeX to do text-formatting from within Sketch. That's very
handy -- the time involved in getting out to edit the text
in a separate TeX session, running it on the commandline, and
importing the resulting EPS file was very *slow*. Doing
it with the script from inside the program is a lot faster.
> What I'm primarly after is to create a _specification_ for a
> wysiwyg-dtp-system, so that programmers later can actually implement the
> system.
My experience has been that such an approach rarely works
out. The open-source world strongly favors a "bottom-up"
design strategy: one finds what is available already and
integrates it into a solution. The alternate "top-down"
strategy of defining what you want, building a framework,
and then filling in the spaces doesn't work very well.
It's even worse when you are just doing the design and
expecting someone else to actually write code for you.
The trouble is that people don't like to work on stuff
that's broken. It's much more fun to start with something
that already works and make it do more. Furthermore,
if they do want to write something new, they don't
want somebody else's spec getting in their hair -- if
you create something on your own, without pay, you're
darned well going to do it your own way!
The top-down approach works okay when you have a small,
devoted group of people at one location (or otherwise
under-thumb) who you can motivate to achieve your goal
in some other way (either you can pay them, or there's
some outside goal that they all want to achieve). The
Gnu project gets away with this *sometimes*.
I don't mean to discourage you -- I just have run
into this problem myself. I think you have to think
about the development process differently if you
want to see any results.
So, given this problem, I would say: look at what's
available already and figure out what additional
features you need for DTP. Then that becomes your
spec. In your shoes, I would select Sketch as a basis,
and try to add the following:
* make native text formatting smarter -- add a
block text object (this is already an interest
area for other Sketch developers, so you may
be able to use what they've done by the time
you get started).
* figure out how to add connected text objects so
that text can flow from block to block.
* add support for multi-page output.
That probably ought to do it -- I'm not sure what
else you'd want that isn't already available. Of
course, making the text formatting smarter can go
a long way: you can just get it to flow ordinary
block text, but there's all kinds of fancy stuff,
like kerning, adding equations, and so on. One
strategy might be to improve the integration with
TeX: instead of just inserting an EPS generated
from TeX, you could write a back-end for TeX that
actually produces Sketch drawing elements (using
all the fonts available to Sketch for example)
and allows more flexible formatting. That's vague,
of course, you'd have to do a lot of research
to figure out just how to fit that in.
Best of luck in any case.
--
------------------------------------------------------
Terry Hancock
[EMAIL PROTECTED]
Anansi Spaceworks
http://www.anansispaceworks.com
P.O. Box 60583
Pasadena, CA 91116-6583
------------------------------------------------------
