World: r3wp

Terry, what you should probably realise is, that that are only probably 
TWO libraries for vector graphics in the open-source world - one 
being AGG, second one being Cairo, library which was selected by 
Mozilla foundation, Apple etc. to do SVG. Both are very good work 
imo ...

And looking at Laszlo - it is nice but definitely slower than VID. 
I think that AGG, performance wise, is even better than Flash ...

http://www.epsitec.ch/cresus/documents/base-f.php- All in AGG! Other 
companies using AGG here - http://www.antigrain.com/customers/index.html

AGG is definitely a win for Rebol (compare it to old draw, right?) 
and it provides us with professional grade vector graphics .... Look 
at Canvas RPaint - very good example of what is it capable of ...

the only problem right now is SHOW, which seems to slow things down 
a lot... but I think I've mentioned that before :-)

How do you test? The actual drawing is done inside 'show, the other 
stuff sets only some values. except you use to-image, that has to 
render too.

well, I'm not really sure what goes on, but if I use cyphres rebcode 
demo it runs fine in a small window, but if I set up the display 
size to twice or three times the original size, the framerate really 
slows down. The drawing routine itself shouldn't care about the resolution, 
so I think it's something in SHOW

SHOW does some kind of "dumb" full screen refresh. it's also visible 
when using Canvas. even if you only draw one pixel, drawing is slow 
in large screen areas.

The following code demonstrates this. Comment / uncomment the draw 
effect line to see the difference a simple draw command makes.

count: 100

do-test: func [label info /local img-size start end] [
	img-size: to-pair label/text
	img: to image! layout [

  origin 0 box img-size blue reform ["Testing" img-size ". . ."] font 
  [size: 24]
	]
	view/new/options center-face layout [
		origin 0 i: image img img/size
		;effect [draw [circle 100x100 100]]
	] [no-title no-border]
	;	test loop
	start: now/time/precise
	loop count [show i]
	end: now/time/precise
	;	display result
	info/text: form to-integer count / (to-decimal end - start)
	show info
	unview
]

view layout [
	title "Resolutions"
	across la: label 100 "800x600" ia: info 100 center
	return lb: label 100 "1024x768" ib: info 100 center
	return lc: label 100 "1280x1024" ic: info 100 center
	return btn "Run" [do-test la ia do-test lb ib do-test lc ic]
]

In answer to Cyphre's question, my opinion is that changes that prevent 
(or make difficult) SVG rendering are not desirable. So if the proposed 
change(s) don't hinder SVG rendering them I'm all for them.

Cyphre question :  i think the same ..  gradient is good the way 
it is for pure draw coders (i you have to make an SVG to DRAW translator 
(my case) you only need gradients that take in charge other kind 
of values nothing more...)

Ok, thanks to all of you who responded to my qestion.

Regarding the SHOW performance: I can tell you that this improvement 
is on RT priority list.

And is it on bottom of that list, or in middle or on top of it? ;)

I'd say it is around the middle line *at the moment*.

OK, I'm just curious :)

Can we fold the paper?

:-)

Velker, I think  you can. Send feedback to Carl about this issue. 
Maybe he will increase the priority after enough requests :-)

ups, Velker=Volker

Cyphre: if your question is "should we create a FILL-PEN-PERCENT 
command to not overload FILL-PEN with percent values?" I say YES. 
(but the name is here as an example)

But overwhise I don't really understand what you mean.

DIDEC you get the point ...  the ask is un precise. My case is particular 
a gradient iss that take in charge percent and vertorial orientation 
to feet SVG datas would be a plus. I think gradient the way they 
are yet implement in Draw/AGG are better than the SVG way. SVG is 
huum less natural less close to human way of thinking and in that 
philosophy it feets perfectly with the REBOL .

It'd be nice to have  VIEWPORT command for creating virtual surfaces 
inside AGG/Draw - especially to use it to constrain text for applications 
such as scrolling and spreadsheets..

viewport origin [pair!] size [pair!] contents [block!]

viewport 100x100 300x150 [font myfont text 150x50 "this is a large 
line of text..."]  ;here the text would use the viewport as its origin 
and screen size

here here

There's already clipping commands.

Why not have a viewport which can be of any shape ?

Heh - i never saw the clip command before :)  Shame on me :)

in my svg render i use push to make a drawing layer order and that 
work pretty good

SVG support the element  in a stackable way so it's good to be able 
to reproduce that feature with draw/AGG push

for example  you draw a black cirle and on top of it you put 2 spots 
yellow and a yellow line  SVG keep the track of that and rebol drew/AGG 
is abe re produce it perfectly...

IS ther a good description of AGG somewhere?

From the author: http://antigrain.com/about/index.html

would it be possible to use agg to create kind of Corel Draw? Simply 
to detect objects you draw? If I understand it correctly, currently 
when once you draw, it is laid down to face bitmap ...

The question is here because not of clone attempt of Core Draw, but 
just curious, if screen-painter could be created. Currently we have 
boxes as effect for 'edge, but ...

of course ability to change mouse pointers to reflect you are at 
right position would be handy to have too ....

do you mean: if I draw line if I can detect if I'm over the line?

yes, detect, press mouse button, have it somehow visually distinguished 
(like bezier curves, those little black boxes over the vector, to 
drag it, you know :-)

That's related to Rebol itself, not just AGG. And it's very bad that 
it's not possible.

I was thinking about making simple vector editor, but it's not possible 
without this feature. In Flash you can detect feels over shapes, 
in Rebol we are limited to faces:(

what do you mean by "shape"?

for example line

you mean non rectangular windows directly in rebol view kernel?

some place where is drawn something

but that would not be sufficient for drawing app as well ... dunno 
what such routines are based upon, but it cant be magic, no? :-)

just have to know what is shape and what is (event) transparent:)

but it should be done in Rebol internally. not that I will make images 
from all drawings and check if it's transparent or not under my cursor

IIRC Tao Intent has such capabilities, you can decide upon alpha 
channel even, so your events could be transparent etc ... well, that 
could be coded too, no? you just press mouse, you detect position 
of your press, and decide what is under the mouse - that is the most 
difficult part imo ...

I think cyphre have some simple examples how to do this but... it 
should be in Rebol internally:)