hi,

pushed for time currently, sorry for omission(s).


"ingo" <nomail@nomail> wrote:

thx for the Nim ref.


> ...
> For writing, syntax and usability testing, of a new SDL language one could write
> a small language, parser and SDL1 transpiler. With a subset of SDL a lot can be
> testet. The output can be the simplest form of current POV-Ray SDL with loops
> enrolled etc.

and Bald Eagle wrote:
> At this juncture, we really need a functional flowchart showing exactly HOW
> POV-Ray goes from the SDL in .pov file to the final rendered image.


diagrams and charts, _yes_.  to add.

I do think the "new POV-Ray" should ship with built-in "schizophrenia" ;-).  if
a version number of less than n.n is given, or the version is missing, the job
goes to the old parser/POV-Ray as of 3.8 (?), else it'll be a version indicating
"SDL2" and the new code takes over.

we need to start with the outline of the "backend", the render engine.  aiui, a
(LL)Virtual Machine is (was?) supposed to be "platform".  as the design of the
LLVM firms up we'll get the (first cut at the) "API".

once there's a specialised rendering "API", we can start thinking about language
features for, I hope, a SDL which compiles[*] to the LLVM low-level
instructions.

[*] many scripting languages, including PHP and Tcl, compile "just-in-time".

with a bit of luck almost all of this can (and should) be "transparent" to us
"end-users".


regards, jr.

Post a reply to this message