On 19/06/2011, at 3:48 AM, George Toledo wrote:
On Sat, Jun 18, 2011 at 11:17 AM, Alastair Leith <[email protected]
> wrote:
It would also be *excellent* to see LUA incorporated as a standard
alternative to javascript, ala Paolo Manna's work.
So replace one scripting patch with another scripting patch.
No, no no.
My point it that, if we're talking about "state machines", the
possible arena of discussion within QC is larger than javascript vs.
"built in QC patches". It's advantageous to have more choices
available, and it's even more advantageous for them to be built into
the system as a stock patch.
Yeah fine but the discussion started about advantage of writing
procedural code to do state machines vs. a noodle soup of concurrency
that at some point gets overwhelmingly difficult for me to modify,
expand, debug or refactor. Let alone revisit in a months time!
That's a paradigm shift! I'm not sure on the winning LUA point of
difference, although I did get excited by the patch when I first
checked it out. My main interest was it's perceived speed over JS,
which turned out to be not so much. And I guess OpenCL is there for
that kind of thing now.
Javascriptcore is fairly fast. OpenCL 1.0 is built more for
scenarios that call for parallel processing, and isn't so great for
many activities, though the CL spec is getting updates that are
making it easier to use (outside of QC, as QC is still 1.0, afaik).
Is it JS's perceived flakiness that's the major issue or the fact
that it is line based code unlike most of the self contained QC
patches?
I have no problem with line based code inside of QC; it's like I
said, it's a feature. It's a great one.
I dislike blanket statements of the kind that have been made about
either path. Both have their strengths and weaknesses. For instance,
when one programs with a Logic patch, or Conditional, etc., you've
exposed ports that allow you to change aspects of that patch's state
in realtime, just from using those patches. That can be
advantageous, and make work quick. Performance may be better, or
worse.
Agreed. We're just describing a JS in QC strength, not running an
election campaign ;-)
You seem to be having a dig both ways.
I'm just trying to be realistic. It could use some work. Why can it
pass image and not mesh, for instance? Why not interaction? The
implementation is either abandoned or arbitrary. However, some of
the built in patches, actually can handle those things (while maybe
having their own issues in some contexts).
Does that matter to you if JS is doing what you want it to? Like you
said play to patches' strengths.
I don't see why QC has to be some kind of steam-punk 'purist'
venture, no text-based coding allowed.
Me either.
Good :-)
If someone is bringing in 200 inputs to JS input ports somebody
needs to learn about using QC structures
What does that mean?
What if a user needs those 200 options to be accessible via a GUI?
We can handle 400 GUI buttons with just three inputs MouseX, MouseY,
LMB_Down. We can assign UI logic to the buttons too them like
exclusivity amongst groups, disable when others active too. All in a
very simple JS script.
and perhaps Kineme Structure Maker I would suggest,
Which takes minutes to load, or doesn't load at all, because of a
bug that causes load time to increase with port count.
I load 100 items (as 10x10 element named structure maker patches) at
>= 50fps on the oldest Mac on this list. That's honestly a new one to
me.
for one thing it's probably going to slow the JS patch down
Again, what does that mean? What if a user needs that many controls?
Refer to answer 2 above.
but that's just my assumption that all that polling is going to be
slow.
The actual patch runs 60fps, even rendering a decent amount of
geometry, it's just unstable as far as duplication of macro or
copy&paste goes.
Weird bug I was unaware of.
But hey why not write state-machines in machine code, it's closer to
the metal than nodes even!
Why not have the ability to drop down to machine code? The inability
to drop down is a major weakness of QC. It would be very
advantageous to be able to open up the loaded patches, edit, make
custom versions, or use the code as basis for our own custom patches.
Access to the Apple CIFilter code in stock filters would be nice to
learn from but as we've discussed they've possibly got Obj-C code not
just the kernal code we get to write in CIFilter patch. Dropping down
to C would neat if it meant speed bumps. There's the Obj-C plugin at
quartzcomposer.com but stock is always preferable, I agree with that.
On 18/06/2011, at 4:17 AM, George Toledo wrote:
I don't think there's anything wrong with using the conditional and
logic patches in QC, fwiw, and I don't think that using them has to
create an impossible messy noodle heaven.
When one configures a javascript patch above a certain input count,
the whole qtz becomes unstable and often results in error when
trying to duplicate macro. Sometimes it's more about using the
right golf club. I think the correct path depends on the criteria
of that particular project.
I think one thing that is an impediment in some of the logic/
conditional patch oriented stuff that I see is that sometimes
people lose track of the fact that the chain in QC starts with
"Enable" on a Consumer, and that it can be odd for people to
conceptualize looping sometimes.
I tend to think that the javascript is a crutch in QC for having
sometimes incomplete paradigms for execution of graphs.
It would also be *excellent* to see LUA incorporated as a standard
alternative to javascript, ala Paolo Manna's work.
On Fri, Jun 17, 2011 at 6:52 AM, Charlie Francis <[email protected]
> wrote:
It's posts like this that make me glad I'm on this mailing list!
Personally I haven't been developing in QC for that long, well
about a year, but I don't see that as a long time compared to the
rest of you guys. When I first started I did exactly what Ade
described. Creating an impossibly messy noodle heaven, with Math
patches, Logic Patches, String Compare, Structure Count and
anything else I could find! I'm slowly making the switch to using
JS to create manageable Compositions, but still struggling even
though I come from a strong JS background of Web Development.
It's always best to sit down and think about what you want to
create first, before opening a New Comp and just "going at it". For
your string switcher, I have previously used these patches to
create a fade in fade out effect. Smooth, Fade and the Sample and
Hold patch. Watch for when the string switches, then Smooth the
Fade value to 0, switch on Sampling for the Sample and Hold Patch
and then Sooth the Fade value back to 1.
Regards,
Charlie
On 17 June 2011 11:38, Adrian Ward <[email protected]>
wrote:
I'm just going to chip in and say that building a central state
machine using the JavaScript Patch has been absolutely critical for
us, and this is what sits at the heart of every interactive AV
we've ever made in QC - without it you'll just get messy
unmanageable noodles, no matter how clean you are with your macro
patching and connection routing.
It also helps to enforce an MVC paradigm on your project, which is
a bit of an unusual approach within the QC ecosystem but I'm
convinced is utterly crucial when making anything ambitious.
Quite possibly my darkest, most sinking moments as a developer are
when I see a myriad of XOR, NAND and OR logic patches tangled up
with Math patches, Counters and other logic type things trying to
control disparate elements that would be much easier achieved with
a single cleanly written JS patch that spits out nice neat state
values. People - Embrace the JS! It is your friend!
Ade.
On 17 Jun 2011, at 10:09, Alastair Leith wrote:
I like Achims State machine, here's another approach I made years
ago before my JS was useful.
I've done this sort of thing where I have two registers and
interpolate between them. The registers are the two most recent
items in a queue, so a new item in pushes the registers if that
makes sense. Can't find a composition for that method.
Also I did it for a structure of 3D attitudes/orientations of an
object. I have a comp for this. In this case the queue just track
random index values, again causing the index at the registers to
shift along each time a new one comes in.
It's a juggling act that uses a pulsed timer (LFO sawtooth-ramp-
up) to drive interpolation patches and the queue; here is a demo
composition I dug out (minus the interesting bit that morphs a
cube into a sphere and back).
<Demo transitions between atitudes with a spinning cube.qtz>
<Rotational Positions.plist>
NB The rotation position.plist is an XML file that needs to be in
the same folder as the comp to load.
Best
Alastair
On 17/06/2011, at 5:09 PM, Rick Mann wrote:
I've been doing these on-screen graphics for a web channel that
covers space launches. We show a couple of different countdown
clocks, as well as a block of ascent parameters.
But for the last nine minutes of a shuttle launch, there's not
much to show. I have a couple dozen events that occur at various
times during the count. I want to display each one as it occurs.
An event is just a text string describing the event ("APU Start,"
"Steering Test," etc.).
My custom patch can either output each string on a output port,
or output an array of structures that has the string and the
associated time. The former is easier for me.
How can I crossfade from the last event string to the next,
especially when they come in rapid succession (perhaps more
quickly than the crossfade duration)?
I was doing a similar cross fade between a set of images, and it
was a real pain to build the structure for it.
Thanks for any suggestions,
Rick
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Quartzcomposer-dev mailing list ([email protected]
)
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/quartzcomposer-dev/qc.student.au%40gmail.com
This email sent to [email protected]
The machine does not isolate man from the great problems of nature
but plunges him more deeply into them.
Antoine de Saint-Exupery
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Quartzcomposer-dev mailing list ([email protected]
)
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/quartzcomposer-dev/adrian%40clayinteractive.co.uk
This email sent to [email protected]
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Quartzcomposer-dev mailing list ([email protected]
)
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/quartzcomposer-dev/charlief%40cellcastonline.com
This email sent to [email protected]
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Quartzcomposer-dev mailing list ([email protected]
)
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/quartzcomposer-dev/gtoledo3%40gmail.com
This email sent to [email protected]
--
George Toledo
[email protected]
www.georgetoledo.com
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Quartzcomposer-dev mailing list ([email protected]
)
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/quartzcomposer-dev/qc.student.au%40gmail.com
This email sent to [email protected]
The machine does not isolate man from the great problems of nature
but plunges him more deeply into them.
Antoine de Saint-Exupery
--
George Toledo
[email protected]
www.georgetoledo.com
The machine does not isolate man from the great problems of nature but
plunges him more deeply into them.
Antoine de Saint-Exupery
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Quartzcomposer-dev mailing list ([email protected])
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/quartzcomposer-dev/archive%40mail-archive.com
This email sent to [email protected]
