World: r4wp

The commit to introduce the doc-strings seems to have caused a number 
of tests to fail, a few in the compiler tests but a much bigger number 
in the interpreter tests. For example:

	ok - serialization....................275 / 275
	** - interpreter-serialization*********11 / 275 **

Wow. Even my docs have bugs. I will run test here. Should have done 
that before.

My tests don't test the doc-strings so its definitely not you Gregg 
:-)

I wonder how I can test your doc-strings ?

I've just run the Red tests on both Windows 7 and OS X. Same results 
on both:

** - Red Test Suite******************5331 / 5666 **

Evenetually we can test for missing doc strings and use some for 
reflection test, at the least.

I get the same serialization results here. Great to have so many 
tests Peter, and easy to run.

Does it log results somewhere?

Yes. the log is in Red/quick-test/quick-test.log

Thank Nenad for the tests being easy to run, it's his design.

Excellent. Maybe have the test runner print out the log location 
at the end?

That's a thought.

I've just checked the quick-test docs and see that I forgot to mention 
the log file. I'll certainly need to update the docs.

Oldes: you can't currently as any-type! is not a defined type yet.

Doc-string related bug fixed.

That's great Doc! Only 12 tests fail here now.

There should be only 9 failing, 3 are invalid tests that I'm fixing 
right now, 6 are EXIT/RETURN tests for interpreter (not supported 
yet).

Yes. Looking good.

Pushed the fixes for invalid tests, you should have only 6 failing 
now.

OpenCV 2D lines demo: http://www.wuala.com/fjouen/Code/OpenCV/Red/pub/lines2.jpg/


Webcam driver and videos playback code example in Red/System can 
be found in this thread: http://www.digicamsoft.com/cgi-bin/rebelBB.cgi?thread=%3C17Mar2013190850930275100%3E


François is close to 100% covering of OpenCV for Red/System. Once 
that (huge) work done, we would need to create some good dialect 
to access all those features from Red.

Awesome!

Dialect =  parse, for that, you need objects, no? :-)

or do you mean Red/System kind of dialect?

Dialects implementations do not need PARSE, they can be implemented 
with base functions, as shown by Kaj with its VID-like dialect for 
GTK+.

PARSE helps making the dialect implementation shorter, faster (for 
interpreters) and more elegant.

weeee wanna more elegant :-)

well, but as for me, from usability standpoint, as a priority, I 
prefer I/O ....

Me too. :-)

float

floats for Red? Yes, why not ... but file, networking, ports, schemes, 
tasking schemes ... so that Graham can implement/port FTP, smtp, 
pop3 etc for us :-)

Pekr, are you aware of this:

red>> 2.3
*** Error: word has no value!


Sorry, but if there is any top priority, than it's decimal support!

Anyway, you can read/write including networking with Kaj's cURL binding 
already, but not to do something basic like 1.1 + 2.2 yet

I don't care about Curl, I regard it being an interim solution. Any 
REBOL-like language has to support what makes REBOL being a REBOL 
in a first place. And one of its concepts is abstracted interfacing 
- ports, schemes, that it :-)

I did not say anything against float support. Actually I said - Yes. 
But my (most probably limited) understanding is, that float support 
for Red is easier than ports etc., so that it will be done, and IO 
comes next ...

If decimal support would be so easy, it would be here already:)

Well, decimal support is in Red/System already, no? To some extend. 
Of course I don't know, if it makes bringing decimal support for 
Red any easier, maybe not ....

extend = extent ....

Right, adding basic float support to Red is not difficult, but as 
floats are not needed internally to build Red, they are low priority 
(but if someone wants to contribute it, it will be welcome). Moreover, 
the runtime lexer is disposable code, it will be soon replaced by 
a new one with Unicode support and more complete syntax support. 
So extending it now for additional literal forms is a bit of waste 
of time.


If someone is interested in implementing float support anyway, the 
decimal! name is reserved for a future BCD datatype, so possible 
names are: real! or float!. It will be a 64-bit float, so mapped 
underneath to Red/System float! type. A support for float32! at Red 
level is not planned, converting float! to float32! at Red/System 
level when needed (i.e. OpenGL API) should be enough.

More screenshots of OpenCV binding progress: http://www.wuala.com/fjouen/Code/OpenCV/Red/pub/

I've started on some Red mezzanines in %Red/mezz/ here. Early days, 
but my hat is off to Doc for how far Red has come already. Yes, I 
crash the console a lot, and I can't DO files yet, but I can paste 
code into the console, play, and still make really good progress.

Doc will have final say over things, particularly what he needs in 
place to make Red self-hosting, but certainly post suggestions for 
your most-used mezzanines.

This is just playground stuff right now, not on github anywhere yet.

Great initiative Gregg! Hope that more people will join.

BTW, Doc, since you did BIND yesterday, I added COLLECT. Need to 
test more, but it works!

HELP and WHAT are next on my list.

Thanks, great addition!

I tried to compile the OpenCV test script for Linux-ARM on a Windows 
system, and I get the following error:

*** Compilation Error: invalid import specification at: libopencv

Is this because I need OpenCV installed on Windows first?

Look in %opencv.reds, the mapping of libopencv for Linux is missing. 
If you can add one there it should compile.

The compiler doesn't need that the imported libs be installed locally 
for the compilation process to work.

The Red/System imports are statically linked into the executables, 
they are just declarations.

linked
 is a bad choice there, "encoded" will be more accurate.