Wed Mar 12 16:39:29 CET 2003
On Wed, 12 Mar 2003, Peter-Henry Mander wrote:
> I think the appeal is in being able to generate elaborate, presentation
> quality documents directly from changable/dynamic data, with charts,
> graphs, tables, links etc...
Actually I started writing it because am *appalled* at the quality
of modern typesetting. IMHO desktop publishing has destroyed high
quality typesetting - for example Times Roman was *designed* for
setting narrow columns of newspaper print - it is commonly used to
print wide columns on A4 paper - the result is a mess.
TeX/LaTeX also are flawed - TeX can't line up rows in two parallel
Just look at LaTeX papers done with two columns - you'll see that
the text in the individual lines are usually misaligned. The use of
LaTeX in producing scientific documents has resulted in virtually all
publications looking identical - the typography is often awful as are
the fonts - believe me there are some really *beautiful* type 1
postscript fonts which deserve to get used in scientific publications.
Many magazines get this wrongs - newspapers are much better.
If you look at a newspaper the overall design is based on a smallish
number of templates - where each template is a set of rectangular
boxes (I guess most papers use less than 20 templates) - indeed there
should only be a small number of templates if the work as a whole is
to have a consistent look and feel.
Erlguten design will reflect this - the idea (not yet implemented)
is to have a small number of templates + a large database of articles.
Making a document consists then of choosing a template and dropping
an article into it.
I imagine generating these (content -> template) map on the fly and
producing personalized content.
The other point is *quality* I want to achieved the highest possibly
quality - MS word etc. get character kerning *wrong* and miss out
ligatures - they make typographically crazy decisions - setting up
fixed columns (a la newspaper) is terribly difficult.
The next level of programs quark-express - InDesign2 etc. can
produce better quality output but are literally *painful* to use (it
probably takes ten thousand mouse clicks to no anything - by which
time your arm wants to drop off).
There seems to be no good free software to produce high quality PDF
from XML (I looked) the industry "standard" way of doing this is to
use XSLT to transform XML to the FOSSI XML DTD and then transform this
to a subset of TeX and then transform this to PDF. This method is
extremely complicated and incredibly difficult to produce any good
results (I've tried).
The direct transformation of XML -> PDF *in one step* is much better
and easier to program - all the Adobe documentation is freely available
and is of surprisingly high quality (well done Adobe) - IMHO PDF is
extremely well designed as is Postscript.
> I intend to use it to generate printable reports directly from and
> within a product testing framework and is accessible from a web page.
Now make me happy by getting the typography right - choose
*beautiful* typefaces - and get the layout right - try reading a few
books on typography first :-) ((seriously)) - things got better and
better from 1445 until about 1985 and then got worse ...
I read that a typesetter setting newsprint on a Linotype machine
could enter properly formatted text 4 times faster than with a modern
desk top publishing system. This was *not* WYSIWYG but all done single
key type setting commands ... If anybody knows how to get a Linotype
manual can they mail me :-)
So let's try to make things better ...
> Thanks Joe, btw, for doing this. I've now got all the tools (pdf,web) I
> need in Erlang, and in the form of tutorials no less. I owe you quite a
> few beers :-)
Great - which bar??
> HP Wei wrote:
> >>Very nice indeed.
> >>So far I have been using Latex for my wordprocessing.
> >>It will be nice to use erlang instead !
> > I just want to know the reason for this.
> > Is erlguten going to be 'Latex' plus more ??
> > hp
More information about the erlang-questions