On Monday 30 June 2008 04:00 pm, Thomas Zander wrote:
> On Monday 30. June 2008 17:31:15 David Gerard wrote:
> > In this case it's something definable. There should be at least a
> > definition possible, outlining what's possible and not and why.
>
> Oh, I can tell you exactly why its impossible. Much like a theoretical
> physicist can tell you exactly why a perpetual motion machine is not
> possible, or why zero calories icecream is impossible.
I finally spotted something that I have some experience with. And I'm replying
to Thomas Zander because he can tell me if I've missed the mark. I believe
this is on topic, because it relates to defining the scope of the effort for
the TC.
ISTM that we haven't defined either pixel or perfect in this context. We just
want the best we can get.
Donald Knuth solved this problem long ago when he wrote TeX by creating .DVI
files that were "DeVice Independent". They were interpreted for each device
that they were displayed on, whether that was a printer, a typesetter or a
monitor.
TeX looked great when it came out. I was doing medium scale publishing then,
sending digital files to the print house. As the printer couldn't handle DVI,
I was stuck - the printer still needed to convert my digital files. I've
never NEEDED to use DVI at home. My equipment just doesn't justify the extra
effort and expense.
ODF applications use pixelated font files, including X fonts and anti-aliased
fonts, scaled for a specific resolution. Even anti-aliased fonts are aliased.
They are just better than normal fonts if used at the right resolution.
If we want perfect, then we need to drop pixels and focus on scalable raster
graphics. But we are working with pixelated fonts, so that won't do at all.
Let's drop the quest for perfection, and specify the tolerances that we can
live with.
So now that we know that pixel perfect and display perfect are not
appropriate, perhaps we can drop back, and figure out just what is good
enough for our purposes. We are on the edge of turning a word processor into
a typesetting system, and that isn't needed.
What quality do we need to have to be satisfied ? We know that there will be
anomalies. Someone is going to send a document that uses a font that the
receiver doesn't have. PDF gets around this by embedding the fonts in the
document. That isn't a good solution for a word processor.
Right now, I just need a decent word processor that I can exchange files with
friends with and be able to print and modify their files.
Current situation is that I can't easily do that. Files don't page right, they
don't stay within the margins, they break lines differently, and on and on.
All I really want from a word processor is to get close enough for government
work. Get the big stuff right, and I don't care if the font is shaped
differently, or if the kerning isn't right, or some other minutiae are
rendered improperly. And if the pagination is correct, we can essentially
reset the error accumulation on each page.
The job before us should be defining tolerances in the right range. It
reminds me of an engineer who transferred from a crane company to a valve
company. He continued to wear his 25' tape measure, causing much mirth
wherever he went. He didn't realize that he had switched from an industry
with 0.25" tolerances to one with 0.002" tolerances. His tape measure was
worse than useless.
Let's get the right scale here.
My list of must haves:
correct pagination
correct margins and tabs
identical line breaks
proper scaling - e. g. lines or blocks don't overlap
font substitution is appropriate and doesn't throw off pagination
bullets don't turn into an empty rectangle
