Writing Extensions

[hh]

>> Alejandro T. wrote:
>> How fast is LCB working with imagedata?

Slower than slowest LC Script ever seen. I Stopped
all my experiments with imagedata after one full day
(was 12 hours).
For a first own test simply take an image of screensize
and walk through the pixeldata for each byte applying
only the identity mapping ...
We have to wait until the java FFI is comfortable enough
to use a java image library. (I couldn't get to there). 

Would be great, if not too difficult to realise, to have
a "do as javascript" here (which uses the js engine only
of the browser widget). This is at least ten times faster
with imagedata than fastest LC Script. See several recent
examples in "Sample stacks" (and some more links in the
Raspi-Subforum to extern servers).

>> Alejandro T. wrote:
>> How fast is LCB working with transform matrices?

Pretty fast and comfortable for transforming all 'small'
canvas data (pathes, text, pattern, image-as-a-whole).
I liked that after taking the first hurdles.

You may test yourself with snippets or lcb files from the
LCB-subforum. There is a lot of stuff for starting there.
Or use the examples on github by Trevor DeVore or the ones
by the LC-stuff, variants by BerndN.

But I wait for my next examples until there is a stable
widget format (we have meanwhile three formats that work on
some LC versions only, not on the others).
To adjust these costs too much time, simply recompiling
worked here only in 1 of 20 cases.