DRAW 1280, 1024

[Benjohn Barnes]

20 years ago, the subject line would have drawn me a line across one
of my screen's diagonals. Ruby is awesomely more powerful than BBC
BASIC, but why no trivial graphics support?

My two requirements are that it should be really easy to draw, and it
should be possible to do so interactively (from irb).

I've looked unsuccessfully in the past, but if this is a solved
problem then I'd really like to know. Otherwise, does anyone have a
suggestion of where to start?

Personally, I'm on a Mac. If there is a solution for my world, that
would do for me. Something global would rock though.

Hope you have a lovely weekend,
Benjohn

 

[Benjohn Barnes]

Because there is no trivial way to make graphics facilities
platform-independent. You may notice that Ruby has native libraries
for
graphic interfaces, but these are also not platform-independent.
Through
those graphic interfaces, you can make drawings, but you then
cannot run
the program anywhere but on the original platform (with some
exceptions).

It may not be trivial, but it's hardly difficult. There must be
dozens of Ruby modules that have varying implementation by platform
(sockets, for example)? Ruby itself pulls the feat of quite nicely
too. Note that I'm not talking about taking full control of graphics
hardware and exploiting all features. It's just really basic drawing
support that would be good.

Well, "irb" has a special purpose, and being a user interface for a
drawing
program isn't the purpose.

I'm not suggesting that irb needs changing. I'm suggesting that the
design of the graphics API I'm hoping for should be "irb friendly",
which I believe it will be if well designed. Thus, using irb, I
should be able to issue a commands and see the result. I shouldn't
have to create contexts and windows, program up call backs, cache my
drawing commands somewhere ready for a re-paint and string the lot
together.

Global, there's the rub. I am sure there are libraries on the Mac
that allow
graphic interfaces to Ruby programs, but they aren't likely to be
portable.

I've not even found something that merely works on the Mac, yet alone
globally!

I apologise if I sound exasperated Paul, and thank you for replying.
It's not like I'm asking for a flying car here though: I just want my
bicycle back.

 

[Chad Perrin]

20 years ago, the subject line would have drawn me a line across one
of my screen's diagonals. Ruby is awesomely more powerful than BBC
BASIC, but why no trivial graphics support?

My two requirements are that it should be really easy to draw, and it
should be possible to do so interactively (from irb).

I've looked unsuccessfully in the past, but if this is a solved
problem then I'd really like to know. Otherwise, does anyone have a
suggestion of where to start?

Personally, I'm on a Mac. If there is a solution for my world, that
would do for me. Something global would rock though.

If you're willing to try a different language, there's always Logo. In
particular, I like UCBLogo (which, unlike other Logo versions,
implements macros along with the rest of the Lispishness of the
language). It has a very friendly functional syntax, has an excellent
lineup of native graphics procedures (aka "functions), and is the
example language used for an excellent trilogy of computer science books
that are available free, online.

Since this is a bit off-topic for this venue, I'll shut up now, but let
me know if you want more information (links to resources, et cetera).

As for Ruby . . . I don't know of any capabilities such as those you
suggest, and I don't see them arising any time soon. I could be
mistaken, however.

 

[Martin DeMello]

20 years ago, the subject line would have drawn me a line across one
of my screen's diagonals. Ruby is awesomely more powerful than BBC
BASIC, but why no trivial graphics support?

My two requirements are that it should be really easy to draw, and it
should be possible to do so interactively (from irb).

You could build something atop Rubygame [ http://rubygame.seul.org/ ]
fairly easily. In fact you could probably write a BBC graphics
emulator, though the cursor-based thing would seem slightly odd in
this day and age A QBasic graphics emulator otoh might be a
worthwhile effort - IIRC it was pretty well thought-out and simple to
use. Here's a quick example using plain Rubygame:

irb(main):001:0> require 'rubygems'
=> true
irb(main):002:0> require 'rubygame'
=> false
irb(main):003:0> Rubygame.init
=> nil
irb(main):004:0> screen = Rubygame::Screen.set_mode([640, 480])
=> #<Rubygame::Screen:0xb78ebc20>
irb(main):005:0> red = [255, 0, 0]
=> [255, 0, 0]
irb(main):006:0> Rubygame:raw.line(screen, [0, 0], [639, 479], red)
=> #<Rubygame::Screen:0xb78ebc20>
irb(main):007:0> screen.update
=> #<Rubygame::Screen:0xb78ebc20>

The simplest way would be to have a class that automatically
initialized a screen and held it as an instance variable, and a set of
methods that wrapped the methods in Rubygame:raw but passed in the
default screen and remembered the last colour that was used in case no
colour was supplied. For example
-------------------------------------------------------------------------------------------
require 'rubygems'
require 'rubygame'

class Sketchpad
def initialize(x,y)
@screen = Rubygame::Screen.set_mode([x, y])
@color = [255, 255, 255] #white
Thread.new {
loop {
@screen.update
sleep 0.01
}
}
end

def line(x1, y1, x2, y2, color=@color)
@color = color
Rubygame:raw.line(@screen, [x1, y1], [x2, y2], @color)
@screen.update
end
end
-------------------------------------------------------------------------------------------
irb(main):001:0> require 'sketchpad.rb'
=> true
irb(main):002:0> s = Sketchpad.new(640, 480)
=> #<Sketchpad:0xb791b808 @color=[255, 255, 255],
@screen=#<Rubygame::Screen:0xb791b7cc>>
irb(main):003:0> s.line(0, 0, 639, 479)
=> #<Rubygame::Screen:0xb791b7cc>
irb(main):004:0> s.line(0, 479, 639, 0, [0, 0, 255])
=> #<Rubygame::Screen:0xb791b7cc>

martin

 

[Martin DeMello]

I'm not suggesting that irb needs changing. I'm suggesting that the
design of the graphics API I'm hoping for should be "irb friendly",
which I believe it will be if well designed. Thus, using irb, I
should be able to issue a commands and see the result. I shouldn't
have to create contexts and windows, program up call backs, cache my
drawing commands somewhere ready for a re-paint and string the lot
together.

There's no getting around the "have to create contexts and windows"
part - base ruby has absolutely no built-in idea of a graphics
context. However, you can do this automatically as follows: to the
bottom of sketchpad.rb add

@sketchpad = Sketchpad.new(640, 480)

def line(*args)
@sketchpad.line(*args)
end

(that is, outside the class definition)

now 'require sketchpad.rb' will execute the lines, create a @sketchpad
object accessible from within irb, and let you call line as a function
straight from the prompt. As a final touch, you can automatically
require sketchpad.rb from your irbrc.

Look also at the Delegator and Forwardable modules for some ways of
automating the method wrapping.

martin

 

[Chad Perrin]

Chad
maybe you remember the post where I was discarding D and you kind of made me
think about it twice. This is a good time to come back [ I consider this
group flexible enough not to be bothered by slighly off topic posts but feel
free to tell me if I am wrong ] to you on this point.
I do not regret at all looking at D a second time and it indeed seems a very
nice thing to have in your toolbox!!!.
So why not toss in the links for UCBLogo?
I feel it might be of interest for more folks than just OP and myself.
If you feel differently about it could you reply to me offlist?
Thx in advance

Good point(s). I'll post links here, at least:

central UCBLogo resources page:
http://www.cs.berkeley.edu/~bh/logo.html

Computer Science Logo Style
Volume 1: Symbolic Computing
http://www.cs.berkeley.edu/~bh/v1-toc2.html

Computer Science Logo Style
Volume 2: Advanced Techniques
http://www.cs.berkeley.edu/~bh/v2-toc2.html

Computer Science Logo Style
Volume 3: Beyond Programming
http://www.cs.berkeley.edu/~bh/v3-toc2.html

At least some Linux distributions (Debian included) provide UCBLogo
packages in their software archives. I was delighted to find that I
could install UCBLogo on my Debian systems by typing "apt-get install
ucblogo" (without the quotes) into a root shell. Quite convenient.

Logo has been called the "Lisp without parentheses", and that's pretty
much what it is. Most implementations have been somewhat neutered in
the creators' zeal for producing a language implementation that is
friendly to young students, but UCBLogo is not of that ilk -- it goes so
far as to include support for Lisp macros (though I haven't personally
explored their use enough to be able to make a determination about how
well they're supported, the syntax and use seems quite similar to that
of ANSI Common Lisp).

It's an open source distribution of Logo created by a Berkeley
professor, Bryan Harvey, who is also the author of the above-mentioned
university computer science textbooks that make use of UCBLogo as the
example language. I've found the books to be well-written (so far -- I
have been often and thoroughly distracted by the demands of other
languages in my life, alas), and the language absurdly easy to learn.
It has been the single biggest factor in my growing appreciation for
arithmetic prefix notation, and functional syntax in general, so far.

Suddenly, I'm considering creating a mailing list . . . damn. Like I
don't have enough on my plate already.

 

[David Vallner]

Personally, I'm on a Mac. If there is a solution for my world, that
would do for me. Something global would rock though.

The best I can think of in portability is using wxWidgets / FOX / Gtk to
draw inside a 2D buffer in a window / GUI widget. OpenGL or SDL to draw
full-screen. I don't think it's trivially easy to doodle over someone's
GUI though - though OpenGL or SDL might be able to draw on the overlay
too. Lookup bindings to those libraries, I'm fairly sure that they're
both available, and reasonably cross-platform.

David Vallner

 

[Martin DeMello]

The best I can think of in portability is using wxWidgets / FOX / Gtk to
draw inside a 2D buffer in a window / GUI widget.

I think all the GUI toolkits insist on running their own mainloop,
which plays badly with IRB

OpenGL or SDL to draw full-screen.

You can run OpenGL or SDL inside their own windows. I think this is
the way to go.

I don't think it's trivially easy to doodle over someone's
GUI though - though OpenGL or SDL might be able to draw on the overlay
too. Lookup bindings to those libraries, I'm fairly sure that they're
both available, and reasonably cross-platform.

Doodling over a GUI is of limited use anyway. Simply having a canvas
you can interactively draw on from IRB is a neat thing. It was one of
the features I'd planned for FXIrb, but I don't have the time to
maintain that these days

martin

 

[David Vallner]

I think all the GUI toolkits insist on running their own mainloop,
which plays badly with IRB

Spawn off inna thread, then jump through event posting hoops?

Just ranting, SDL would be prolly both easier to get to work and with a
richer API for drawing.

 

[Martin DeMello]

Martin DeMello wrote:

But (IMHO) you shouldn't be using irb anyway. You should be using a user
entry event loop that evaluates the user's entries with "eval".


I am at a loss to understand why you think irb is essential to your plan.
There are much better solutions for talking to the user.

Because IRB has already solved the general problem of a good, robust
REPL. Why should drawing to a canvas be treated as a separate problem?
Ideally it should just be another set of things you can do from the
ruby REPL.

martin

 

[Chad Perrin]

Because IRB has already solved the general problem of a good, robust
REPL. Why should drawing to a canvas be treated as a separate problem?
Ideally it should just be another set of things you can do from the
ruby REPL.

The upshot, I think, is that you'd probably have better luck creating a
new REPL interface than integrating what you want to accomplish with
IRB.

 

[Martin DeMello]

Because of the problem of the event loop, which will be hijacked by irb if
it's part of the application.

That only applies to a full-fledged GUI, complete with listeners,
though - all the OP wanted was a simple canvas to which he could draw.
I've already posted a working proof-of-concept of a canvas with a
thread to update itself every few milliseconds, controllable from irb
- I see no reason why that can't be extended to a more featureful
drawable object that works happily alongside irb.

martin

 

[M. Edward (Ed) Borasky]

20 years ago, the subject line would have drawn me a line across one of
my screen's diagonals. Ruby is awesomely more powerful than BBC BASIC,
but why no trivial graphics support?

My two requirements are that it should be really easy to draw, and it
should be possible to do so interactively (from irb).

I've looked unsuccessfully in the past, but if this is a solved problem
then I'd really like to know. Otherwise, does anyone have a suggestion
of where to start?

Personally, I'm on a Mac. If there is a solution for my world, that
would do for me. Something global would rock though.

Hope you have a lovely weekend,
Benjohn

Well ... the closest to "global" that you're likely to get is Ruby/Tk,
which is a port of Perl/Tk. There are some hints about getting started
with it in the Pickaxe book, but you'll end up wanting a good Perl/Tk
book to learn everything that it can do. As far as global is concerned,
it at least runs on Windows, Mac and Linux. I'm sure it works on AIX,
Tru64 and Solaris and can be made to work on HP-UX .

Another source of inspiration/guidance is to install a full Tcl/Tk
distribution. It's available on all the major Linux distributions and
from ActiveState (www.activestate.com) for Windows. It's probably
available for MacOS if not installed by default. The worst case is that
you need to build it from source.

Inside the Tcl/Tk distribution is a rather nifty collection of demos --
really cool things like drag and drop, sketchpads, etc. The source code
is in the Tcl scripting language, but is easily translated into Perl,
Python or Ruby.

If you're really adventurous, you might try Rails and Ajax. Most of the
common GUI things, including drag and drop, are available in libraries.

Hmmmm ... there's an interesting Ruby quiz/ smallish educational project
... write a translator that converts the Tcl scripts in the Tcl/Tk demo
package to Ruby/Tk.

 

[Logan Capaldo]

Good point(s). I'll post links here, at least:

central UCBLogo resources page:
http://www.cs.berkeley.edu/~bh/logo.html

Computer Science Logo Style
Volume 1: Symbolic Computing
http://www.cs.berkeley.edu/~bh/v1-toc2.html

Computer Science Logo Style
Volume 2: Advanced Techniques
http://www.cs.berkeley.edu/~bh/v2-toc2.html

Computer Science Logo Style
Volume 3: Beyond Programming
http://www.cs.berkeley.edu/~bh/v3-toc2.html

(Getting _really_ OT now...)

I just want to say thanks, the chapter on Automata was really good,
and it had the best explanation of the proof of the halting problem
I've ever read.

 

[Chad Perrin]

[ snip ]

(Getting _really_ OT now...)

I just want to say thanks, the chapter on Automata was really good,
and it had the best explanation of the proof of the halting problem
I've ever read.

Quite welcome. I'm glad it was valuable.

Now if I can just get Brian Harvey to relicense UCBLogo under less
restrictive distribution terms than GPL, my work will be complete.

Sorta.

 

[Benjohn Barnes]

Thanks very much for all of the posts - I'll read them all on the bus
in a moment!

Cheers,
Benjohn

 

[E. Mark Ping]

Now if I can just get Brian Harvey to relicense UCBLogo under less
restrictive distribution terms than GPL, my work will be complete.

I'd say that's unlikely. He's quite the idealist (he was my first
instructor in CS, and I was a TA for him later). Not quite RMS (he's
far more relaxed), but in the same vein.

On the other hand, if you offer him enough potstickers, you never know
what he might do. A BSD-style license might interest him.

 

[Chad Perrin]

I'd say that's unlikely. He's quite the idealist (he was my first
instructor in CS, and I was a TA for him later). Not quite RMS (he's
far more relaxed), but in the same vein.

On the other hand, if you offer him enough potstickers, you never know
what he might do. A BSD-style license might interest him.

Idealism isn't an obstacle to me -- it's the wrong ideology. The things
I dislike about the GPL have nothing to do with commercializing things
as proprietary software, and everything to do with granting people the
right to distribute software freely. The GPL, in case you haven't
heard, has been used by the FSF as justification for threatening small
Linux distros with lawsuits if they don't devote significant resources
to maintaining, and making available, archives of software for several
years after the software in question has become obsolete by the
project's standards. According to the GPL, once you make a piece of GPL
software availabl to someone in binary form, even if only for a few
hours, you must then provide easy accessibility to the source for three
years thereafter -- and linking them upstream is not sufficient to suit
the FSF.

This is going to make a lot of would-be Linux distributors think twice
about creating and distributing a distribution. It will especially
affect smaller project would-be founders think twice.

It made me think twice. I now won't ever be part of any GPL binary
distribution chain, ever -- which pretty much makes BitTorrent a
non-option for me as a means of getting Linux installer images, since
BitTorrent's value for that is negated when I'm not allowing uploads to
"give back" to the torrent availability. A distribution I was working
on has been canned. Et cetera.

 

[M. Edward (Ed) Borasky]

Idealism isn't an obstacle to me -- it's the wrong ideology. The things
I dislike about the GPL have nothing to do with commercializing things
as proprietary software, and everything to do with granting people the
right to distribute software freely. The GPL, in case you haven't
heard, has been used by the FSF as justification for threatening small
Linux distros with lawsuits if they don't devote significant resources
to maintaining, and making available, archives of software for several
years after the software in question has become obsolete by the
project's standards. According to the GPL, once you make a piece of GPL
software availabl to someone in binary form, even if only for a few
hours, you must then provide easy accessibility to the source for three
years thereafter -- and linking them upstream is not sufficient to suit
the FSF.

This is going to make a lot of would-be Linux distributors think twice
about creating and distributing a distribution. It will especially
affect smaller project would-be founders think twice.

I can't personally imagine why someone would want to be a "Linux
distributor". There are two major "for-profit" Linux distributors
already, Red Hat and Novell, three major community distributors,
Slackware, Debian and Gentoo, and lots of little splinters and hangers
on. If you're motivated by money, work for Red Hat or Novell or someone
who uses them. If you're motivated by community, contribute to Debian or
Gentoo or Slackware.

There are a lot of things not to like about the GPL and the FSF in
addition to the ones you mention. But even in the absence of potential
FSF lawsuits, there are lots more interesting and potentially profitable
problems to be solved within the context of existing distributions,
rather than creating a new one.

It made me think twice. I now won't ever be part of any GPL binary
distribution chain, ever -- which pretty much makes BitTorrent a
non-option for me as a means of getting Linux installer images, since
BitTorrent's value for that is negated when I'm not allowing uploads to
"give back" to the torrent availability. A distribution I was working
on has been canned. Et cetera.

I wasn't aware of the BitTorrent interpretation of this FSF policy. Has
someone actually been asked not to participate in BitTorrent exchange of
Linux install media by the FSF? I download Quantian, Gentoo, CentOS,
OpenSuse and occasionally other distro images all the time with BitTorrent.

Speaking of licensing, a couple of days ago I downloaded OpenSuse 10.1
and installed it. You ought to read the Novell EULA when you install it
-- I almost bailed out and probably won't use it, since it was quite a
bit slower than Gentoo anyhow. Oops ... one of the terms of the EULA
was that I not publish benchmarks.

 

[Chad Perrin]

I wasn't aware of the BitTorrent interpretation of this FSF policy. Has
someone actually been asked not to participate in BitTorrent exchange of
Linux install media by the FSF? I download Quantian, Gentoo, CentOS,
OpenSuse and occasionally other distro images all the time with BitTorrent.

Nobody associated with BitTorrent distribution of GPL software binaries
has been threatened with a lawsuit or otherwise coerced to cease
participation in BitTorrent exchanges YET, but I don't want to run that
risk personally. Someone was the first person to get individually sued
by the RIAA -- someone may be the first person to get individually sued
by the FSF. I'll let it be you.

Speaking of licensing, a couple of days ago I downloaded OpenSuse 10.1
and installed it. You ought to read the Novell EULA when you install it
-- I almost bailed out and probably won't use it, since it was quite a
bit slower than Gentoo anyhow. Oops ... one of the terms of the EULA
was that I not publish benchmarks.

I'm still waiting to see if EULAs are even enforceable -- that hasn't
been settled yet. In any case, I don't favor SuSE of any flavor, so
that's not much of a concern for me personally. That doesn't mean I
approve of the EULA approach to things. Retroactive contractual terms,
particularly when implicitly applied, don't make me happy in the least.

You may notice that I snipped all your rambling to the effect of "Nobody
should want to create a new distro anyway." There's a good reason for
that: it's not for you to decide what someone else does or does not want
to do, and it is in large part the ability to fork and/or compete as a
lone enthusiast that keeps the FLOSS community vibrant. All of your
commentary about why one might or might not want to spin off another
distribution is irrelevant to the point at hand, as far as I'm
concerned.