AltME groups: search

Kaj: let me know how it performs on your test machine with Mandelbrot.

First test looks three times as fast. I'll shut down the machine 
to do a clean test

Okay, libc-init is now working fine for Linux and conforming to Gcc 
ABI. Once again, man has won over the machine, but I wonder how many 
neurons were killed both in my and Andreas brain during the battle...;-)

harr..harr. andreas the human coding machine :))

Red is actually quite suitable for CGI, because the compiled executables 
are absolutely tiny, like they were in languages such as C when CGI 
was introduced. This means the performance will be quite good, not 
like the performance hit CGI suffers with heavy modern programming 
environments such as REBOL where the whole virtual machine needs 
to be started for every request. At the same time, you still have 
the advantages that made CGI popular: simplicity, robustness and 
security

Recently tried to run the Hello World on my Android based Galaxy 
Samsung Tab 10 but after seeing the "executing ..." msg. Nothing 
appeared. I will try to look at what could be done but I suspect 
I also have to first learn a bit more about Android and my machine 
to help ... I'm very patient and began to read about the ARM. and 
Android itself. What an evolving world we live in ...

I will provide the info tomorrow when I have access to the machine.

I've added information about my machine to #219 (no gcc info, as 
it doesn't have gcc installed).

Can anybody check this code? https://gist.github.com/3135678

It's not a bug, but I wonder why the obviously more complex sine-osc 
is cca 50% faster than square-osc. These are the results I get on 
my machine:

sine-osc time:  1068
square-osc time:  1790

I think the latter was Windows, so that would make ik very susceptible 
to anything else going on in the machine

No, I don't think so. The Red runtime is written in Red/System. Red/System 
compiles to machine code.

Robert, Nenad's intial plan was for the Red compiler to directly 
generate machine code. So his first presentation reflects that. It 
was a few months after the launch that he decided to generate Red/System 
code from the Red compiler.

I just checked on my old machine, I did re-wrtie the lib tests on 
OS X 10.6 but on OS X 10.7.

My virtual XP machine is running I think Belgian localistation (turned 
off, but the base system is localised). W7 host is Czech localisation. 
I should try this with original US version.

One hypothetical advantage you have with Red is that you can make 
the logical behavior fairly high-level and have the compiler/optimizer 
get rid of that at runtime. REBOL, being interpreted, is effectively 
a lower-level language requiring hand optimization, the kind of hand 
optimization that you'd want to prohibit in Red because it would 
interfere with the machine optimization. This means that, for strings 
at least, it would make sense to have the logical model have a lot 
of the same constraints as that of R3 (because those constraints 
were inherent in the design of Unicode), but make the compiler aware 
of the model so it can translate things to a much lower level. If 
you break the logical model though, you remove the power the compiler 
has to optimize things.

Inline ASM: we might add that at some point, but you can already 
access some low-level CPU/FPU features (like stack manipulation). 
I will extend that in the future to support CPU in/out commands and 
direct registers reading/writing, so ASM support will be less necessary. 
Also, there's a cheap way to support it, adding the option to inline 
machine code directly in Red/System code flow (you would have to 
use an external ASM for generating the machine code). Something like: 
inline #{....machine code...}

Thanks. I know, I'm low on glitz. It's because I don't want to take 
the preparation of those talks more time than they already do. I 
prepare by making sure that as much as possible works, and then I 
do a guided tour of it on the machine itself

OpenCL: yes, I have that in mind too, but it's most probably for 
a later version of Red (> v1.0). Also I have other plans to give 
Red access to GPGPU, by directly having a GPU machine code assembler 
and a dialect on top of it. It would be mainly for compiler internal 
use, but such dialect could be exposed for user code if some are 
willing to implement some fast GPU-powered routines.

If you're doubting about such approach, read this article:

 http://www.geeks3d.com/20110317/low-level-gpu-programming-the-future-of-game-development-on-pc/

I'm sitting next to a Windows 7 Ultimate machine to test the bindings, 
but it's unusable because it has been updating itself unasked for 
hours, on shutdown and on startup, and now won't start up anymore

An Internet connection for a Windows machine is close to a lethal 
injection

well, that's the first thing I do when I set up a Windows machine

It's not my machine, and safe mode or F8 doesn't provide an option 
to restore the configuration

Petr, this machine just became a ghost

All that talk about my machine, I touched one button and it awoke 
again ;-)

We have new Samsung 64bit Win8 machine in the office, I might try 
next week to run some tests there too ...

I had to kick AVG from the Windows 7 machine to be able to use Red 
and thus Red in Fossil

OK, I just installed win 8 Pro 32 on one of my machines and I downloaded 
the v0.3.0-0 zip and ran the run-all.r file. Perhaps that's not the 
right thing to do since it printed "running lexer-test" and then 
proceeded to lock up my machine. I can't even ctrl-alt-del. Let me 
know how you would recommend me testing Red since I may be the first 
to have win 8 among us but the last to really use Red passed the 
original "let's see if I can see 'hello world' on my Droid" a few 
months back :-)

One-based is human friendly, while zero-based is usually more machine 
friendly, so I think REBOL made the right choice

Well, I thought, that Red code gets compiled/translated into R/S 
code, and that code is going to be translated into machine code :-)

Yes, and that doesn't magically make it being programmed in machine 
code

I fixed the compile bug in World, so Kaj's Fibonacci test now takes 
8.7s to run in World on my machine. R2 takes 17s and C less than 
1s.

On my machine, REBOL catches up enormously when it can do many iterations 
in a very small loop. Apparently, a very small function will eventually 
fit in the CPU cache, so the VM gets less of a cache hit compared 
to compiled languages

Does call/console work under Windows 7?


call/console "ver" causes REBOL to hang on my machine after printing 
"The handle is invalid twice."

As far as the wish for a View/VID native solution goes, I wish that 
as well. Maybe it will be possible when the JIT compiler becomes 
reality to easily adapt Rebol's VID. 

It does not have to be complete like VID, a basic set of widgets 
will get you a long way.  

Other solutions are a really good thing too, but looking at GTK and 
myself I find it hard to find out how to get GTK on my mac. It is 
not a standard dmg file I can download and install and it works. 
Other GUI solutions require integration of their package or having 
the end user of your programs to find out how to get it running on 
their machine. That kind of thing can be a real showstopper to global 
acceptance.


I know Doc is working hard and has already stated he intends to come 
up with a VID like native solution. So we can let him focus and be 
silent ,or we can comment and discuss letting him know we do care.

It would be really nice to have automatic builds+tests running on 
all major supported platforms for each new commit on main branch, 
and an automatic report generated online. That would really saves 
us time and avoid missing some obvious regressions.


I know that Andreas started working on that. Andreas, do you already 
have a good plan to achieve it?  Do you need help/resources?


I can provide a Linux server for Linux/x86 targets. Having a Windows 
machine online + MacOS X would be great. Also a Linux/ARM (RPi for 
example) that is always up and reachable from the net could nicely 
complete the list.

I'm feeling nostalgic.  I used to hand-craft machine code on the 
6502/6510 series of processors back in the 80's, so I have an idea 
of what you're talking about.  There are some pretty neat tricks 
you can do when you're dealing with memory addresses directly.  I 
crafted more than one self-modifying program to squeeze every bit 
of computing power out of a processor.  I know that a lot of that 
ability has been removed by protected memory and abstraction, though.

Thanks for the report, Gregg. It could also be that this XP machine 
doesn't have networking. Perhaps Windows doesn't understand that 
it can use localhost

Livecode cross-compiles to any platform.  I can build Windows, Mac, 
Linux, Android, and Web apps, and X-code projects, with the click 
of a button *all on my Windows machine (or on a Mac or Linux box, 
if I want).

In any case, we just bought an XP machine for testing

(On my machine, current constellation of the stars, etcetera, etcetera)

Well - IDE is IDE, I mean - the runtime stuff. Simply put (and most 
probably worrying unnecessarily and sounding like a broken machine), 
let's make R3/Red another Amiga like OS :)

I'm moving to a new dual-core machine. When compiling Red, it uses 
only one core, so I'm considering using 0MQ to make use of the extra 
core

It's only seven years old. :-) My previous machine is eleven years 
old

But seriously why not a machine that is more actual? Because you 
prove the minimalistic needs?

That is to say, the old machine was already more than usually needed 
for Syllable

Well, looking into binary .class, I expected it being some kind of 
executable or package, just pressing Enter does nothing :-) So it 
is not mapped to JRE on my machine .... I selected it manually, and 
it says:

Could not find or load main class, or something like that ...

The first class files where generated on Andreas 32-bit machine.

(My .class files where generated on a 64-bit machine, but that does 
not matter.)

Hi Doc, congratulations for this new landmark. As the hello.apk generated 
in the past from Red/System was not running for my Galaxy tab 10.1 
was not running, I expected the new apk release not running too and 
it is effectively not running here opn this machine, even  if I upgraded 
to Android 4.0.4 recently - so the problem is located under the Android 
level - as I had expected it too. May be the op cpodes generated 
are not the ones expected on my machine - that is they are not currently 
compatioble with the ones you generated - but I have no time to test 
it. this was just to keep you informed so others know this too... 
Don't bother with this case as I plan to get another Android device 
in the next few months to compare the compatibility of the product. 
Probably the CPU codes are not athe level you deliver them (ARM v5 
if I remember correctly ?)  Best regards.

@Doc : I just sent back my unfiltered log results for the Samsung 
Galaxy tab 10.1 machine - thanks and best regards

@Arnold and Pekr : What is the current behaviour of the Hello.apk 
? I thought it was displaying some welcome msg and then askin for 
some input before disappearing gracefully. Now on my machine a while 
screen is displayed with a "Hello" as its window name and nothing 
else is going on.... For the previous Red/System Hello, something 
almost similar was going on too. So it's not for the Red part fault 
but it is due to the lower layer ... I knew this but I can't diagnose 
what ios coming myself as I'm missing some vital knowlwdge here to 
be really helpful to debug anything ... on my own - sorry for having 
acquired some product in the first pioneers just to test how this 
kind of product was running on Android. I required some Flash and 
my Apple products just could'nt do it - but my tablet Android tablet 
was much slower on display than my iDevices. Hope that under Red 
and Red/System there will be some products running at full speed 
- if not I plan to let it go at a real bargain price very soon ... 
to get some more powerful device ASAP.

Now on my machine a while screen is displayed with a 
Hello" as its window name and nothing else is going on...."


Gerard, that's exactly what the current hello.apk is supposed to 
do.

Quick testing of the MT Red/S and C versions on my machine give a 
~4x difference in speed.

On my machine, it loads but then comes back with error 1058

I was mislead by an incorrect intermediate result in the Red/System 
source that I expected to be correct already but had not fixed on 
this machine, together with a short blackout overseeing that adding 
nonzerobits to the left of a binary number changes the decimal representation 
in the lower digits too. Lots of output lines, many output terminal 
windows, small laptopscreen.
Same results make sense.

Normally I do not announce my upload to rebol.org but this time is 
different. I have uploaded an upgraded version of  my interface to 
opengl: http://www.rebol.org/view-script.r?script=opengl-glu-glut-h.r
and I am asking you to insert "GLX_DOUBLEBUFFER" in the block of 
the line: "attrList-def: reduce [GLX_USE_GL GLX_RGBA  GLX_DEPTH_SIZE 
16 GLX_RED_SIZE 4 GLX_GREEN_SIZE 4 GLX_BLUE_SIZE 4 0]"  to see if 
doublebuffer works since in my virtual machine it is not working 
(an empty window is opened and no gl rendering occurs)

Normally I do not announce my upload to rebol.org but this time is 
different. I have uploaded an upgraded version of  my interface to 
opengl: http://www.rebol.org/view-script.r?script=opengl-glu-glut-h.r
and I am asking you to insert "GLX_DOUBLEBUFFER" in the block of 
the line: "attrList-def: reduce [GLX_USE_GL GLX_RGBA  GLX_DEPTH_SIZE 
16 GLX_RED_SIZE 4 GLX_GREEN_SIZE 4 GLX_BLUE_SIZE 4 0]"  to see if 
doublebuffer works since in my virtual machine it is not working 
(an empty window is opened and no gl rendering occurs)

I haven't tested it much yet, because I haven't emulated a machine 
around it yet

In this new emulator, I abstracted the machine emulation from the 
CPU emulator, so that different machines could be implemented using 
the same CPU emulator

I've tried several operating system ROMs. They seem to work, although 
it's still hard to judge without a real machine emulated around the 
CPU

Thanks to a build machine provided by HostileFork, OSX PPC ("2.4") 
builds of R3 (from mainline Git) are now available for your convenience:

http://www.rebolsource.net/

(Look for "OSX PPC" under the "experimental builds" section.)

This work will be paid for by Respectech as part of an assignment 
I'm doing for them. As far as we know, this is the first commercial 
assignment done with Red.


I'm implementing a network server that needs to run for long periods 
and handle relatively large numbers of requests. This is tricky because 
Red has no garbage collector yet, and the memory allocator doesn't 
handle large series yet. We have been testing the memory system to 
scale it, and Doc has fixed an allocator bug so that series can now 
be twice as large. I've also found and reported seven other bugs 
so far.


The results are very encouraging. The memory system is reliable in 
my stress testing. It can use all the memory available in a machine 
for the contents of series. In the default configuration, one series 
can have a maximum content of 2 MB, or 131070 values for a block!. 
This can be enlarged, basically at will, by compiling your program 
with customised allocator settings.


To reduce the amount of memory that currently cannot be reclaimed 
by garbage collection, I am reducing the number of series generated 
during the handling of each server request. This can be done within 
the same Red program, because low level optimised Red/System code 
can be inserted anywhere. I am down to only a few hundred bytes per 
handled request, and the memory budget of the server is a GB, so 
the Red 0MQ server will be able to handle several million requests.

New World alpha release at
https://github.com/Geomol/World


- Added better networking using cURL library, libcurl (OS X and Linux)
- Added tasks incl.:
	- task! datatype
	- task-id! datatype
	- TASKS native function
	- task? and task-id? mezzanines

 - Task support for many functions incl. HELP, SOURCE, FIRST, SECOND, 
 PICK, COMPILED? and DISASM
	- KILL mezzanine to call TASKS/KILL

- Added inter-task communication using SEND and RECEIVE native functions
- Added support for messages to WAIT as: wait 'message

- Added preemptive multitasking using TASKS/TICKS to specify number 
of Virtual Machine instructions per task run (default is 200)

- Added timers for tasks triggered by WAIT native and READ native, 
when reading from url
- Task yield can be achieved by: wait 0
- Changed ;{ } comment syntax to ;( )
- Parens can now span several lines at the prompt
- Added system/console/paren
- Added support for pressing <Esc> to stop execution of main task

- Sending input via pipe to World started with a script will activate 
input without echo and without <tab>-completion etc.. (Not sure how 
and if this works under Windows.)
- Added support for picking 'Re and 'Im of complex numbers
- Better implementation of TRY

- New test versions of IF and EITHER, that can take other than block! 
arguments
- Added more tests

I think I know the source of this problem! The script did work on 
my machine without ANY error!
Why?

I see no difference between Java and Rebol. Both free to use on every 
machine the company builds a VM capable of processing scripts/prgrams.

You mean in the video? That's a faster machine

Currently it only works on Windows, and you need the dyn-lib-emitter 
branch of Red/System
 I did n't know you have a Windows machine ;)

On several occasions in my life, I've programmed on a machine I don't 
have

Pekr, he already did port it to ARM Linux, but at the time noone 
working on the project had an ARM machine to test with other than 
an Android phone. And porting the core to Android is the simple part; 
the host would need to be rewritten almost completely for Android 
due to the different application model. Noone was using Amiga either. 
The only alternate platform that was getting any use by the project 
members was Syllable.

Pekr, he already did port it to ARM Linux, but at the time noone 
working on the project had an ARM machine to test with other than 
an Android phone.


Maybe I fall into the "no one working on the project" category, but 
I had ARM machines available for at least the last 6 years or so. 
Last time Carl was rumored to have ported to ARM (most likely "again"), 
I was certainly around, with ARM machines, offering to test :)

It seems mostly meant for testing RebCode. It lacks a lot of functionality 
needed for emulating a real machine

I rigged up my C64 with the speech synth module so that it would 
read out each character as i typed in the machine code listings etc

Saphirion contributes the OSX machine, for which I'm very grateful.

Thanks, Doc. After fighting for a day and a night with the XP machine 
to keep it from self-destructing, I could test that audio playing 
also works on XP now - but no task bar icon...

IIRC you can install the virtual loopback adapter in order to allow 
tcp on a card-less machine on XP.

indeed.   adding an actually stable and simple virtual drive system 
to windows would be nice.  I've tried two free ones, and they just 
end up crashing the machine when used.

Could anyone tell, Is rebol a language for sketching systems, or 
is it a full-fledged Turing-full language with access to all machine's 
resources? Seems to be a sketching one, IMHO. Could anyone comment 
on this?

REBOL is structured more like LISP or Haskell (but without being 
a pure functional language).


So, yes, it does not have direct access to the machine instructions 
or low-level op sys APIs in the way that assembler or assembler-wrapper 
languages (like C) does.


What REBOL does have is easy integrated access to very high level 
APIs: parse. bind, map, etc.

Ladislav, some questions are still open. I am currently remotely 
connected to my machine. I'll study your "lesson" tomorrow and I'll 
reply.

The chess program I want to make give the moves of a piece like on 
the shredder site http://www.shredderchess.com/daily-chess-puzzle.html
and it will have a minimal validation routine so my kids (and me 
too) can beat the machine.

Where does open-proto come from in the mysql-protocol.r in the mysql 
driver from softinnov? Word has no value in the version I am using. 
Strange. No complaints when using the script from a location on my 
webspace, trouble when used to connect from my local machine to the 
web MySQL.

if you are looking for a fultext/searchengine don't miss elasticsearch. 
scales well, based on lucene, very fast, simple to use.

the interesting part is the compact rest api the storage model and 
the number of different analysers, tokenizer (snowball, ngram, etc). 

you can use elasticsearch also as a standalone nosql database with 
rebol. speaks completly json. 

working currently with 1.2 m documents + datastorage on a single 
machine.  runs very smooth. 

evaluated other components before (sphinx, xapian, tokyo etc.) elasticsearch 
has it all and is much cleaner to handle than solr.

A 1 GHz machine is very powerful for Syllable, but it sounds like 
you may be hitting the bugs in the USB 2 driver. Do you have USB 
1 ports that you could try?

Kaj, not sure if it's of use to you, but I dug out the timings from 
the last time I built Qt (on a Linux host, 4 months ago):


$ ./configure -opensource -nomake examples -nomake demos -nomake 
docs -nomake translations
5 minutes

$ make -j2
34 minutes


That was on a moderate dual-core machine with 2GB ram and spinning 
disks. Not sure how much make could parallelise, but for a single-core 
machine roughly twice the build time will be a reasonable worst case 
estimation. I guess that otherwise the configure-to-build time ratio 
will still be roughly the same.

I'm starting to get the GUI build to work, and it takes a few hours 
here, but Syllable Desktop can't use the second core on this machine, 
and compiling is roughly half the speed as on Linux, anyway, so that's 
roughly consistent with your timing

To start some new discussions, 1) I like Squeak/Pharo Smalltalk's 
AIDA/Seaside systems. Hope to to practice them in realworld soon 
(creating a community site)... 2) I'm coding a new Oberon system 
 which will be able to do one of the following at once: a) interpret 
oberon b) translate oberon to x86 machine code c) translate oberon 
to javascript. This is a very long-term project, but I get excellent 
support from the oberoncore.ru community, and am actually making 
good progress. Current oberon project status is c++ sourcecode-based 
ssystem interprets oberon, interpreter has many temporary stubs.

But I have "crashes: 11" for RT-built A111 on this machine as well.

I think the additional guess is another real regression on the machine 
I'm using. But let me find out which test is newly crashing.

Running on a misalignment-tolerant machine, the unoptimised 64b binary 
is actually slightly faster than the unoptimised 32b binary.

Outlook and Thunderbird can "just" look up what your local machine 
claims to have configured as hostname.

I don't have an ARM machine available to test

otherwise, he could probably make make on a different machine, where 
he can get an r3 interpreter

Some exmplanation:


W->top_of_series is a stack holding the different types of series 
being entered in the lexer, defined as:

#define BLOCK_BEGIN_T		58
#define PAREN_BEGIN_T		59
#define LONG_STRING_BEGIN_T	60
#define PATH_BEGIN_T		61
#define GET_PATH_BEGIN_T	62
#define LIT_PATH_BEGIN_T	63
#define BINARY_BEGIN_T		64
#define SET_GET_PATH_T		65


The code about trace is just, if tracing is on or off in World. W->top_of_code 
is a pointer to where the code for the virtual machine in World is 
being created, and W->code is the bottom of that stack in instructions, 
W->top the top. do_string executes a string of World code.

with higher resolution displays, comes higher performant CPUs/GPUs. 
Tegra 4 announced, 4core A15 1.5 GHz + 1 A8 Core, 72 GPU cores .... 
I still think, that we did not utilise full power of AGG yet. Gee, 
View 1.x ran on my Pentium 75 machine ....

or virtual machine code
or real machine code

Also chasing svm.r Simple Virtual Machine which was related to the 
KOSH project

Oh and I have some fx5 styles stored in a backup of an old REBOL/View 
cache that is not on wayback machine.

The upload obviously worked -- "+" is not an illegal file name character 
on either Edgars machine or REBOL.org's current host

The notes on contributing a packgae do have detailed guidelines on 
file naming to avoind problems -- see 2.1 here:

http://www.rebol.org/cgi-bin/cgiwrap/rebol/boiler.r?display=cons-upload-package-files-help.html

But, of course, the author probably wrote the application before 
they had any idea it might be rolled out as a package for use on 
a wide range of platforms.
***

The bottom line is that + ought to work in a file name for the downloader, 
so I'll get it fixed.

If I can throw my two cents here. Terry is right to complain, because 
you can feel he cares, it's not destructive criticism. Pekr is saying 
he's a realist, but to me he sounds more like a dreamer (no offense).Flash 
is a Virtual Machine that lives in the browser. Rebol is a VM that 
lives on several OSes. Flash VM is present on 99% of the browsers. 
Rebol VM is not present  on browsers (plugin doesn't count because 
last time I tried it wouldn't install, and last version is more than 
a year old), and it is present on OSes of a small number of creative 
developers and hobbyists. If I had to develop a commercial app for 
the web, Flash and Flex give me extensive documentation, and a whole 
framework. Rebol can't compete with Adobe, and should not. Last week, 
I developed a GUI in Rebol and Rebgui that would have taken me four 
times longer in any other graphical toolkit, and I had a knack at 
it. But this app is for internal use. Rebol is a great language, 
and I would like to see it evolve as a computer language, not as 
a throw it all in kind of gizmo. It should go Open Source, because 
RT doesn't have the means to provide versions for all platforms. 
Where is the rebol for NetBSD? Last version is 2001. If RT hasn't 
the resources, let the developers do it. I want to run rebol on an 
internet tablet. I had wanted to run rebol on a palm (not anymore 
because Palm is deas, sort of). But the Nokia tablet run debian linux. 
There is no reason for the unabailability of rebol. It is just a 
matter of building it and packaging it. Who has the time for this? 
Plenty of people, they're just not at RT, because there people are 
busy with more important things (like developing the product).

Funny enough, I will be posting soon two articles: "Why I don't use 
the Mac" and "Why I don't use Linux"  and I was going to open this 
up to everyone to do battle with me on each point.  The idea being 
that all can learn from this and make the move to the machine they 
like best.  For some people they might read one of the articles and 
say "those five things don't matter to me, so I'm going to buy a 
Mac now" 


For others, it will serve as a thermometer of when it is worth going 
to Mac or Linux.