There is now a new web page at http://object-handles-examples.appspot.com/ that has 7 example applications intended as either a training resource, or as a starting point for more complex applications.  I’m charging a fee for the source code to those examples.  The fee is whatever you decide to pay (with a $15 minimum)

The more revenue that this generates, the more motivation I’ll have to spend time working on the core library and additional examples.

The core library will continue to remain free and under the MIT license.  It can always be downloaded from http://code.google.com/p/flex-object-handles/

There were a few reasons for me to do this including:

1. My motivation to work on the library has been very low lately.  I don’t have any current projects that are using it.
2. The donations link that I have has received a total of 3 donations over 2 and a half years.
3. By working on these examples, I was able to really exercise some of the core functionality.  While doing this I was able to

• Completely revamp the constraints mechanism and implement things like an Aspect Ratio constraint
• Find and fix a handful of bugs
• Figure out the answers to some questions others have asked that I just didn’t know the answer to
• Make several big performance improvements

P.S.  This gave me a chance to create a real site in Python/Django/Google App Engine for the first time, and I’m impressed.

This is for an order placed 1/18, and I have no confidence that it would have actually gotten here on the 25th so I’m canceling that.  It wouldn’t be so bad if the people you have to talk to on the phone had some idea about what’s going on.  Something like “We just can’t get that video card” or “The wireless adapter is out of stock”.  I would have modified the order to get it sooner if possible.

But all I got was corporate-speak about how great dell was and how I’d just have to wait.

I’m buying a mac mini instead.  At least apple will screw me over with a smile.

Daniel has been kind enough to offer me a small percentage of their sales as a thank-you for the ObjectHandles library.  It’s MIT Licensed, so he didn’t have to do that, but I’d like to send a big thank-you to him for that.  That’s a great way to support open source projects.

Do you enjoy playing Fraction Nation?

I don’t enjoy playing Fraction Nation.  I Love it!  Very Inspiring!  You guys probably change my life and others!  Wonderful Job!

Or this one from another kid…

If you could change something about the program, what would you change?

The thing I would change would be nothing ’cause I really liked it!

Another kid, same question:

If I were to change something, I would change the background.

Oddly enough, several kids said the background thing.  I say if the background is the biggest thing they want to change, that’s a big win.

Universally in that class, the feedback was overwhelmingly positive.  Making a product that teaches fractions that kids like isn’t very easy to do, but I think we might have nailed this one.

And here’s a doodle one of them made on the back of a survey for us.

The thing that really scares me is almost every one of these kids has better handwriting than I do now.

It’s not working 100%, and I’m not sure why. If anyone has any ideas, let me know. The broken part is if you rotate 2 objects, then select them both, then resize them really small. At some point they sometimes start getting bigger instead of smaller depending on the rotation. I’m baffled.

The second thing I played with is the idea of “Decorators” that can draw interesting info based on the state of the set of objects either being moved or currently selected. There’s a quick screenshot of a sample one I did below.

This is also in the link above under “Example 8″. This is just a proof of concept for now. If you want to try it, do it before trying the multi-select since the multi-select can leave objects on fractional pixel boundries which won’t line up perfectly.

Neither of these is ready for prime-time yet. The multi-select stuff is creating tons and tons of temporary objects so the GC kicks in periodically freezing the interface. That’ll need a little optimization.

http://code.google.com/p/flex-object-handles/downloads/list

I hope to write some better documentation on how to use this new release soon. If you’ve been using ObjectHandles Version 1, it’s not a straightforward port to the new stuff. There are drastic changes to how it all works.

Problem #1

When working in Flex Builder, if you have a lot of non-embedded assets, it can go really slow.  At seemingly random times Eclipse will decide it needs to delete all of your output files and re-copy them.  For our current project, that can take 10 minutes.  We have yet to figure out what triggers this behavior.

Problem #2

When working in Flex builder, if you have a lot of embedded assets, it can go really slow.  Especially on clean builds, or builds that touch a lot of modules.  This is an every-compile type thing, but is far less severe than problem #1.

Solution for Problem #1

Don’t let Eclipse do the copying for you.  Put your assets outside of your source tree, and turn off “Copy non-embedded assets”.  To the right is an example directory structure.  There are a few differences than most Flex/Flash builder projects.

1. My assets folder is a sibling of my src folder, instead of inside it.
2. My compiler is set to output to assets/bin-debug instead of just bin-debug

In your source, you’ll have to link to your assets a little bit differently.  Here’s an example from my TestProject.mxml that shows both an embedded and a runtime loaded asset.

Since the embed is relative to the source folder, we need to use paths in the form of ../assets/art/*

The runtime loaded assets are relative to the final compiled location, so those take the form ../art/*

Now, eclipse will never try to manage your assets.  This appears to completely solve problem #1.

Solution for Problem #2

A while back I blogged about turning a Flash Catalyst project into a .swc.  (Read that here)

I dusted off my script and tried it with the Beta-2 of Flash Builder and Catalyst.  And low-and-behold it worked without any changes.  The only thing I had to do was reset my FLEX_HOME environment variable to the new location.

So, if we were to move away from a model where we embed assets from designers into Flash Builder, and to a model where the Designer puts those assets into Catalyst, and then we run a script to compile the catalyst project into a .swc, I think that would completely solve problem #2.

I had been using one of the G1 phones on TMobile’s network.  It was running Android 1.6, the new Droid is running Android 2.0.  Let’s compare Droid to that first.

Let me start by saying, they’re not all that dissimilar.  The one plus the G1 had was the keyboard was easier to use.  It actually had separate physical buttons whereas the Droid has them all on a single etched piece of rubbery plastic.  The droid is faster, has more memory and storage, and looks better – all by a lot.  Holding the phone if feels like it’s engineered and manufactured well.  Holding the G1, it kind of felt like a toy.

I haven’t had any problems finding or running any apps.  All the ones I’ve tried have been great.

The new Google Maps with navigation is amazing.  Better than my Garmin stand alone GPS unit.  It’s got “layers” now that can overlay all kinds of things.  Some of my favorites include traffic and Wikipedia.  The Wikipedia layer shows all of the Wikipedia references on a map.  When looking around my home, there’s maybe a dozen points listed.  I’m hoping a Geocaching app takes advantage of that layer functionality soon.

The battery life seems good.  I used the navigation app for an hour and a half on my commute to work one day.  That includes the screen being on the entier time, GPS updates contantly, and lots of network traffic.  As well as the speaker telling me the directions and a traffic overlay showing me where the road would slow down.  At work, I left all of the services (bluetooth, wifi, gps) all day long.  I consciously tried to use it as much as possible.  At the end of the day there was still 15% battery life left.  Normally, I plug it in while driving, but it’s nice to know I dont have to.

So, if those things were the only differences from my old phone to this one, the $200 TMobile cancellation penalty, plus the $200 new phone cost might not have been worth it.  But there was one major other difference.  The network.  Verizon is amazing compared to TMobile.  I’ve used the same speed-testing app on both phones in a variety of situations.  The fastest speed I ever saw with TMobile was around 300kbs.  When I was at my house, it never went over 5kbs, usually in the 1-2 range.  With Verizon, the slowest I’ve seen is around 300kbs.  At my house, I average 500kbs.  That over a 25000% increase.  When in the office at work, it wasn’t worth even trying to get a data signal with TMobile.  With verizon, it works perfectly.  This is the biggest single selling point of the entire experience for me and I’m am a very happy customer.  This speed changed the way I use my phone from the G1 tot he Droid as much as the G1 changed the way I use a phone from a dumb-phone to a smart phone.

Cost-wise the monthly plans between TMobile and Verizon are about the same.  I do get a 17% discount because my company has a contract with Verizon so your price may varry.

I heard a great Droid radio ad today.  This one below is similar, but the Radio one was better.

Problem number one, it’s WAY too easy to schedule a meeting with someone.  Since the bar is lower, more meetings happen.  Since there’s more meetings, they’re often individually less valuable.  Since they’re less valuable people don’t treat them with the respect they deserve.  Nobody prepares beforehand.  Meeting agendas are the exception, not the rule.

Secondly, it’s not an efficient way to schedule a meeting.  Wait, let me revise that slightly.  It’s not an efficient way to schedule a set of meetings.  The normal process that is often used is to put everyone on an invitation.  Find a block of time everyone is available.  Book it.  Later, that process is repeated for the next meeting.  Even later it’s repeated again.  Since different people go to different meetings, individual schedules get “fragmented”.  Back in the days of personal assistants (or back then I guess the term was “secretaries”), they would handle the process of defragmenting schedules to make them make more sense.  This leads to situations when individuals have an hour meeting, an hour to work, an hour meeting, an hour to work, etc.  Without bigger blocks of time to work it’s impossible to tackle certain problems.  Even worse, a high priority meeting might get pushed back because other meetings were scheduled earlier.

Combine those two problems, and it actually gets hard to book a meeting.  Since there are a lot of them.  And since there is fragmentation in people’s schedules, finding an open block that everyone can join can be really hard to do.  This means scheduling happens further and further out.  If you need to talk about something with 10 people and the only time slot to get them all in a room is a week away, that sucks.

A side-effect of this is that when you receive a meeting request, you often really feel obligated to accept it.  Especially if the sender knows you have a free block of time.  That means you’re not asking yourself  ”Is me attending this meeting the best way for me to spend my time to advance/improve/help the organization?”  I think we should be asking that question all the time!

How to define an ideal solution?

If we’re going to explore other options, it’d be helpful to have some evaluation criteria.  Here’s what I would look for in a meeting scheduling system (in my order of priority).

1. Ability to schedule meetings without conflicts.  (including resource and people conflicts)
2. Schedule more important meetings sooner than less important ones.
3. Minimize the number of chunks of free time each user has.  (In other words, maximize the average length of non-meeting time)  This one is really important for knowledge-workers.
4. Encourage productive meetings.   (Clear purpose, set agenda, everyone comes prepared … )

What other properties might an ideal solution have?

So what’s one solution?

First, here’s a technical solution.

The first problem of meetings being too easy to schedule is tough.  It really requires training people to act differently.  Without designing a technical solution that somehow discourages meetings, I’m not sure what to do about that.

The second problem could have a technical solution.  Imagine a “smart” scheduling system.  You need to schedule a meeting?  Submit the details of the meeting to some computer system.  Tell it

1. Who needs to attend
2. Any time constraints that must be upheld.
3. What resources do you need.
4. How important is it?

But the important thing is that you don’t specify when it happens.  (There may be time-constraints for some meetings that do dictate a time.  Like if a client is arriving, it has to happen in a given time slot.  Or if there is a release planned for thursday, the release planning meeting must happen before thursday.)

Then the system is free to schedule that meeting whenever is “best” by some smart criteria.  It tries to maximize individual people’s free time blocks.  It tries to schedule more important meetings sooner.  It moves existing meetings when new ones are scheduled so there’s always an optimal solution planned.

The downside?  Meetings move.  It’s hard to plan ahead.  Someone needs to develop a complex scheduling system.  The time that meetings happen becomes “voodoo magic” to some people.  It’s possible to game the system.  It’s not flexible.

Solution #2

Now, here’s an organizational solution that I want to try out here at TSP.

1) Meetings that HAVE to happen at a certain date/time are scheduled as normal.  Got outsiders coming in?  Meeting to watch the pat’s game?  Go ahead and schedule those as before through outlook.  Hopefully that’s the minority of meetings for you.  (If you’re the guy meeting with outside clients all day long, you’re pretty much screwed any way you look at it.)  A certain class of exceptional or emergency-meeting falls in here too.

2) Got a meeting that doesn’t HAVE to happen at a certain date/time?  You’re not allowed to just schedule that through outlook.  Instead, at your daily scrum we schedule those as a group.  With one more constraint.  You only schedule meetings for tomorrow.  It’s either important enough to schedule for tomorrow, or it isn’t.  That’s a much easier, concrete decision to make than scheduling a meeting next-week.  Some meetings lower on the priority list might not fit.  That’s fine, try to get them on the list again tomorrow.

To actually create this schedule, you’re likely looking at outlook as a group.

Benefits:

1. You can intelligently schedule meetings.  Since you’re only looking at a day at a time and you’re considering all the meetings together it’s easy.
2. You get to decide which meetings are more important than others.
3. You get to weigh the benefits of a meeting against non-meeting things.  It’s easy to schedule a meeting a week in advance for a trivial matter.  It’s really hard to schedule that meeting *tomorrow* if there’s more important things to do.
4. It makes people think about meetings and whether or not they have to happen.
5. Since a meeting you’re invited to is only ever a day away, it’s more on your mind and you might prepare for it instead of being surprised by an outlook reminder.

Down sides:

1. Boy does this require a lot of buy-in from everyone involved.
2. Participants have limited time to prepare (The person calling the meeting has as much time as they need to prepare since they could just wait another day if needed)
3. Since your group is scheduling a day in advance, it can be hard to get ahold of resources that other groups might be scheduling weeks in advance.  There’s likely solutions to that, but I’ll stay away from them for now.

I’d only consider using this for larger (4+ people) meetings.  Still feel free to have impromptu talks whenever they’re needed.  Still feel free to try an resolve issues at the scrum, or by sticking around after your scrum.  In fact it might encourage a quick post-scrum discussion instead of a meeting.

Side-Note: I heard a rumor that Warren Buffet only schedules meetings a day in advance.  I have no idea if if is true or not.  But that’s where I got the idea from.

Caveats

Replace “Outlook” with any other “dumb” scheduling solution out there.  Lotus Notes?  Google Calendar?  Whatever.

Don’t complain that I call it outlook and not Exchange or whatever the entire-system should be called.

My solution assumes you have a short daily meeting.  It would suck to have to introduce that if you don’t.

What else?

There has to be better solutions to this.  What do you do?

At work, whenever a developer submits a change to a project we automatically make a build to make sure nothing goes wrong.  This board gives us the current status of four different automated builds that happen on those cruise control servers.  The green light indicates that the last build was successful (it switches to red if it failed).  The blue light lets us know if it’s currently building (it does a nice pulsing fade in/out).

On any failure, a flag is also risen from the top of the board.

There’s two meters at the top.  The first one shows an overall project completion (that’s manually entered by me right now).  The other one shows “communication density”.  Right now, it’s a random value that ticks up and down.  Eventually it’ll represent the volume of traffic on our project mailing lists.

The top gauge looks like this (The 0-180 scale is a bit of an internal joks that any fellow TSP employee would get):

The electronics are fairly simple.  4 pins of the arduino control the red/green lights.  4 pins control the blue lights.  One of them controls a servo that raises/lowers the flag, and two pins are for the two meters.

Besides that I have a switch on the back that activates “demo mode” that does some pretty flashing lights and raises the flag.  There are a couple status indicator lights that tell how the board is functioning back there as well.

The meters are 1970’s era amp-meters in the micro-amp range that I opened up and replaced the background label with customized print outs.  I was suprised to see that a simple resistor and PWM signal could control them accurately.  I was thinking I’d need a capactitor in there to even out the current.

Here’s the back of the board.  Normally, there is a panel that covers all this to make it look less like a fire hazard.  At the top you can see one of the meters.  Next to that is a small protoboard that the status lights, demo-switch and flag-servo are all soldered into.

If you follow the green wires down from there you’ll come to my Arduino board which is actually a lower-cost Seeduino that I bought so I could continue playing with my arduino for other projects.  One nice thing about the Seeduino is you can solder connectors directly to it.  I’ve got 2 8-pin connectors attached to it that connect all the lines I needed to the proto board below it through those rainbow-wires.  That second proto-board has the resistors for the blue lights and the connections for all of the lights on it.

To the left of all that is a structure that holds all of the lights, the lenses, and some dividers between the lights.  To control two lights with a single pin it was easier for me to put those resistors closer to the LEDs so that’s what you see there.

The black wire from the Seeduino to the foreground is a USB cable providing power and status updates.   On the computer it’s attached to there is an AIR application that queries status info from our cruise control servers and relays it to a program running on the seeduino.

Dominik Pesch over at 11com7 has created and sent me a .mxp package for Pulse Particles. This package adds the particle explorer as a new window within the Flash authoring environment and makes the component available from the standard libraries.  This could really make it easier for designer-types to use.

Download PulseParticlePackage.mxp (4mb)

Thanks Dominik!

1) Plug board into USB
2) Install Arduino software
3) Select example blink sketch (application) and upload

And suddenly you have a blinking LED (one is build onto the board).

So then I plugged in a little servo I had lying around, found the servo library.  Wrote a little sketch, and I was able to make the servo go to any position by communicating with my desktop computer over USB.  Total time, maybe half an hour.  Controlling servos isn’t the easiest thing in the electronics world since you have to worry about timing signals & whatnot.  But Arduino boils all that down to 2 or 3 lines of code.  Wow.

So next I set off to grab as3glue, which includes a serial port/tcp gateway so you can control the arduino through flash.  After maybe 20 minutes of tinkering I have an amazing high level API I can use in flash to control the board. (I ended up using arduino2flash instead of serproxy since serproxy wouldn’t work for me.)

Here’s a little AS3 snippet that uses a utility class I wrote to control 4 LED’s attached to the board from flash.

That code causes each LED to fade in and out over and over. Wow… tweening physical computing? Neat.

I had long been thinking about tinkering with one of these. If you have too, just do it. $30 for the board, and an hour of your time will get something working.

Next up, building a Build Notification system that will read in our CruiseControl status at work and let people know what’s going on.

You just need to add an “implements” attribute to your root tag and specify the full class name. I’ll stick to the base class for large components, but this is the way to go for small, quick things.

It’s still a rough site, but I figure Flex 4 took me about the same amount of time to make it as it would have in Flex 3, but that includes learning about all these new features.

So I fired up the application in the Flex profiler.  Let it sit for a minute, clicked the “Reset Performance Data” button, waited another few seconds, and then clicked the “Capture Performance Data” button.  The result was odd.

Nothing was actually in code I could touch and didn’t explain why it was happening.  There had to be something going on to peg out the CPU like that.  So I hit up the Flex preferences panel and found something I had forgotten about.

By default, the profiler excludes a whole bunch of flex and flash related classes from the profiler.  So I removed those exclusions and unchecked the box.  Went back to my profile window and low and behold, this is what I saw:

So now I had an idea of what was happening.  It looks like that indeterminate progress bar (which wasn’t on the screen anymore by that point) was sucking up some render time.

So I peeked at the ProgressBar flex code.  They’ve got a timer in there that gets started, and is stopped when the progress bar is set to invisible.  But that doesn’t end up accounting for cases where we just bump to a different parent object in a view stack.  So that timer sits there, happily spinning and consuming resources.

I added some code in to manually set the visiblity of the 2 indeterminate progress bars I had and ran.  Bamn, success.

So I learned (or re-learned?) two things today.

1. When profiling, if you don’t see anything happening check your filters.
2. Be very careful with indeterminate progress bars in flex.

Note: I got similar results both in the debugger (ADL) and when running the application stand-alone as a bundled application.