braslau.list at comcast.net
Tue Dec 4 19:54:54 CET 2018
On Tue, 4 Dec 2018 18:03:14 +0100
Hans Hagen <j.hagen at xs4all.nl> wrote:
> As most mechanism are reasonable okay now we can look into what
> future versions of ConteXt should provide in terms of functionality.
> The (luatex) engine is also more of less finished or at least stable
> in terms of functionality.
> One possible area of further development is graphics so I'd like to
> know if there are (reasonable) demands for more graphics support
> using metapost. Alan and I are on-and-off working on some extensions
> and one of the things we discuss is visualization of data. The
> question is of course what should be done in the lua/mp/tex
> combination and what externally but the combination has some
> advantages. There has been presentations and discussions at the last
> meeting and graphics is always a nice topic, and also kind of
> rewarding in terms of development.
> Of course other demands can be formulated too, but these then need to
> come with well defined descriptions (or mockup examples).
Hans is alluding to work that we are doing to explore the possibilities
of lua + MetaPost in the treatment and visualization of data,
efficiently, with the possibility of handling large quantities of data.
Lua does this particularly well.
My first exercise has been a complete rewrite of the John Hobby graph
macros. It was/is an exercise in recreating the functionality "from
scratch", trying to use lua effectively when needed. The code and
approach is entirely new. My use case was the analysis of a large set
of data - entirely ridiculous for graphical display but containing
"outliers" or instances that are significant, and it works! I am
trying to generalize, and am doing this in my "free time".
So the aim is two-fold:
1) to efficiently handle large datasets through the use of lua
2) to generalize the notion of user data space from the
two-dimensional drawing space of MetaPost (and the represented page),
and affine transformations of this 2D space to other coordinate
systems and dimensions. We are not thinking about 3D rendering or
raytracing, but rather simple projections and other transformations
such as polar coordinates. (Of course, log scales, for example are a
"simple" non-linear transformation.)
I guess that Hans would like to expand the discussion to the future of
graphics in general, inviting input from everybody.
More information about the ntg-context