Re: More MetaPost text positioning....

On Sat, 2006-02-25 at 12:10 -0500, Rob McDonald wrote:
A bug report with a some summary of this mail is the right thing to do.
It does not matter much if you reopen the previous one or open a new one.

Ok, I'll do that.

Talking of UML (class?) diagram I had another issue with the matapost
plug-in: collision with TeX escapes and special chars marking the
access of class members. Could you try Self/dia-core.dia and suggest
some solution (maybe escaping as done in pstricks?).

Yeah, I'm talking about the class diagram.  I'm using it for a database
schema, so whatever you want to call it...
Of course, this problem will always be there.  LaTeX does pretty things like
ligature (special merged hybrid characters) on ff and fi.  It also does
kerning (squishing together to remove inter-letter white space) on letter
combinations like VA.  Dia won't ever match this, so very subtle text width
problems will always persist.  Of course, these are really small problems...

Yes, those problems exist with almost all exporters.

A much larger issue is if a user wants to insert a LaTeX formula into a

$E=M\ c^2$

MetaPost export will handle this fine for now, but Dia doesn't have any clue
as to the actual width and height of this text.  Imagine if someone were
entering an array or a large, complex equation with bulky fractions.  (This
is another reason that I think you should always be able to override the
width of a box, even if the text then overflows the sides.  Because, when
LaTeX renders it, its width may be really different.  However, I see this is
a LaTeX only feature, and isn't very important.)

As I see it, we can either want to have full LaTeX rendering in Dia
(i.e. call out to LaTeX to get a rendering) or have the LaTeX output be
as close to what's in the diagram as possible.  That would give the
least surprises.  A nice thing would be to be able to toggle if a given
text should be tex-interpreted or not, but since right now we have no
call-out to LaTeX, we're going to exact rendering as best we can.

For an actual LaTeX rendering, we'd want to be able to give some
context, of course, but if we can get a bitmap from the dvi output
(which is possible), we can kinda treat that as an image in Dia.  Trying
to actually understand and calculate what LaTeX does to our text would
IMHO just create more headaches than solutions.

An example of this kind of functionality is the TeXPoint PowerPoint plugin.
A guy has developed an open-source powerpoint plugin that lets you use LaTeX
as an equation writer (bitmap image generator) for PowerPoint.  Double click
on the image, and a LaTeX text editor opens up.  Change the equation, click
OK, and it seamlessly compiles the LaTeX to a specified resolution bitmap,
and imports it into PowerPoint.  It is pretty slick...  He did it with his
own home-brew LaTeX distribution.  Making a similar technology
cross-platform would be a nightmare.

Exactly what I'm hoping for.  I would be happy to include something like
that (though probably not for homebrew LaTeX only).


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]