Showing posts with label HowTo. Show all posts
Showing posts with label HowTo. Show all posts

Thursday, December 19, 2013

How to get Ref parameter values in RhinoMock

Mocking ref (and out) parameters with RhinoMock is a bit tedious. I ran into something that could possibly be considered a bug. At the very least it's not really expected behavior. The good news is I found a work-around. Hopefully this will help someone else (including perhaps my future self).

The basic problem is: Ref parameter values passed to mocked methods have the default values specified in the mock setup and not the values passed by the calling code.

The background

Take the following interface as an example.
interface ISomething
{
    void DoStuff(string a);
}
Now supposed this is mocked and some things need to be done based on the value of parameter a.

Normally this is easily done:
public void Initialize()
{
     mockSomething = MockRepository.GenerateStub();
     mockSomething.Stub(s => s.DoStuff(Arg.Is.Anything())
          .Do((Action)(arg =>
               {
                    // Perform general action with arg
                    
                    if (arg == "abc")
                    {
                         // Special case action for arg
                    }
               }));
}
Every time something calls mockSomething.DoStuff, the code passed to the Do method will be executed and the parameter arg will contain whatever value was passed to DoStuff. This is routine stuff for RhinoMocks and works as expected.

The setup

Now suppose the parameter for DoStuff was a ref.
interface ISomething
{
    void DoStuff(ref string a);
}
This is where things get a bit dicey. The interface is still mocked and some things still need to be done based on the value of parameter a. So, some minor syntax changes in the arguments constraints to handle the ref stuff for the compiler are done:
private delegate void DoStuffDelegate(ref string a);

public void Initialize()
{
     mockSomething = MockRepository.GenerateStub();
     mockSomething.Stub(s => s.DoStuff(ref Arg.Ref(Is.Anything(), string.Empty).Dummy)
          .Do((DoStuffDelegate)((ref arg) =>
               {
                    // Perform general action with arg
                    
                    if (arg == "abc")
                    {
                         // Special case action for arg
                    }
               }));
}

The problem

The above code compiles. And it runs. It just doesn't run correctly. The problem is that second parameter in the Arg<T>.Ref method. It indicates a return result for the the value. The problem is RhinoMocks sets the parameter value to the return result before calling Do's method. In other words, in this example, arg will always be string.Empty. The code in Do will never be called with the values sent to DoStuff by the original caller.

Looking at the call stack, I could see the original method call with the correct parameter values. Then it went into the RhinoMock and proxy code and then Do's method was called, clearly with the unexpected value.

Looking for solutions

Digging around, I found the WhenCalled method. This appears to be a bit earlier in the mock/proxy processing so I changed the test.
public void Initialize()
{
     mockSomething = MockRepository.GenerateStub();
     mockSomething.Stub(s => s.DoStuff(ref Arg.Ref(Is.Anything(), string.Empty).Dummy)
          .WhenCalled(invocation =>
               {
                    var arg = (string)invocation.Arguments[0];

                    // Perform general action with arg
                    
                    if (arg == "abc")
                    {
                         // Special case action for arg
                    }
               }));
}
Nope. This didn't work either. The value for Arguments[0] has already been set to the return value.

While searching around, I found other people asking about the same issue. In their cases they found alternative solutions based on constraints of when their methods were called and what values the parameters could have. With known values, constants can be used via hard coding. For example, instead of using Is.Anything() as above, Is.Equal("abc") can be used and the second parameter can be "abc". Everything is fine.

I was semi-successful with the special case by using this technique. But then I needed to do the special action and use Is.NotEqual("abc"). I ran into the same problem as with Is.Anything(). I didn't know the original value for arg.

The solution

Widening my search, I stumbled upon an old article by Ayende talking about the Callback method. He considered it for weird edge case parameter checking and indicated it shouldn't generally be used. As far as I could make out from his write-up, it's purpose is as an alternative to the Arg constraints when they aren't sufficient.

Having nothing to lose, I changed my code to give it a try:
private delegate bool DoStuffCallbackDelegate(ref string a);

public void Initialize()
{
     mockSomething = MockRepository.GenerateStub();
     mockSomething.Stub(s => s.DoStuff(ref Arg.Ref(Is.Anything(), string.Empty).Dummy)
          .Callback((DoStuffCallbackDelegate)((ref arg) =>
               {
                    // Perform general action with arg
                    
                    if (arg == "abc")
                    {
                         // Special case action for arg
                    }

                    return true;
               }));
}
Hurrah! This worked!

Since Callback's intent is to determine if the constraint on the stub is valid, it gets the original parameter values, rather than the return result.

And, yes, I realize I'm misusing the intent of Callback. But when nothing else works, you go with what does.

The conclusion

Since this met my needs, I stopped here. If this was a function rather than a method, I suppose a Do or Return method would have to be chained to the setup code after the Callback method in order to return a valid value for the stubbed function. Also note, if this stub should be ignored, then false can be returned from Callback instead of true. This would allow other stubs for the same method to be handled differently.

I'm not sure if this is a bug, intended behavior or an unconsidered edge case that has, at least to some, unexpected behavior. Both Out and Ref methods have the return result. It makes sense to be mandatory for Out, but I think it should be optional for Ref. I can see cases where you'd want to stub in a particular value all the time. The current syntax supports this well. But I can also see cases where it shouldn't be changed, at least by the Ref handling code. An overloaded version of Ref without the parameter would work well. In any case, I don't think it should be set before WhenCalled and Do are are invoked. At a minimum it should be after and better yet only if the original value hasn't been changed by WhenCalled or Do.

Well, that's my latest story regarding RhinoMocks. I hope it helps someone.

Friday, April 12, 2013

Unit Tests Not Found in VS2012 and How It Was Fixed (partially)

I am setting up a virtual machine for a new upcoming project. It will use some core libraries that have already been developed in-house with Visual Studio 2012 as the development environment. The team that developed the existing libraries did a good job to make sure there was a suite of unit tests that worked and was reasonably comprehensive. So, after I installed VS2012 and Resharper I retrieved the code from the repository, compiled it and tried to run the unit tests.

The Resharper test runner (the one I normally use) listed all the tests and reported their status as "Inconclusive. Test not run."

The test runner built into Visual Studio didn't list any tests at all. Zip. Nada. Nothing.

A web search revealed some posts indicating there was a compatibility issue between VS2012 and Resharper that was supposedly fixed in VS2012's Update 2. I downloaded and installed the update.

After this, Resharper listed all the tests and reported their status as "Pending" without doing anything further. A change, but arguably not as good as before. To exit with things pending just seems strange. At least the previous "Inconclusive" state gives a sense that something's wrong, even if it doesn't say what.

On the other hand, Microsoft's test runner... still did the same thing. Absolutely no indication that any tests even existed.

After quite a bit more searching, chanting various incantations and sacrificing not a few livestock, I stumbled upon an ancient forum post from early last summer when VS2012 was still gestating as an RC. It indicated there was a bug with unit tests when they existed on a network share.

Hmm. I wanted my VM to be just the build environment and had placed the source code on a shared folder that was mapped alongside all the other source code on the host machine's disk. I wondered what would happen if I moved the code to the C: drive.

As soon as I loaded the project from its new location on C:, Microsoft's test runner immediately found and ran the tests.

Unfortunately, Resharper's still doesn't work. It continues to set them to "pending" and never does anything else. Apparently this is a known issue that hopefully will be fixed soon.

(Hopefully I've added enough keywords to this that others will be able to find this problem and solution without having to spill blood all over the keyboard.)

Wednesday, October 31, 2012

Sandcastle Help File Builder and error MSB4062 on Team 2008 build agents

I recently had to add a Sandcastle Help File Builder task to our automated build process. Our product consists of both .Net 3.5 projects and native C++ projects. It compiles under Visual Studio 2008 with Team 2008 build agents.

The latest version of Sandcastle is built against .Net 4.0, so, as part of its installation, I also installed the .Net 4.0 runtime. Everything worked fine in the GUI but when the build agent tried to run, it failed with:
error MSB4062: The "SandcastleBuilder.Utils.MSBuild.BuildHelp" task could not be loaded from the assembly C:\Program Files\EWSoftware\Sandcastle Help File Builder\\SandcastleBuilder.Utils.dll. Could not load file or assembly 'file:///C:\Program Files\EWSoftware\Sandcastle Help File Builder\SandcastleBuilder.Utils.dll' or one of its dependencies. This assembly is built by a runtime newer than the currently loaded runtime and cannot be loaded. Confirm that the <UsingTask> declaration is correct, and that the assembly and all its dependencies are available.

The UsingTask declaration appeared correct and all the dependencies seemed to be in order. I searched for quite a while and found several places purporting to have answers, but none of them worked. As I expanded my search, I finally happened upon an InfoSupport article about building VS2010 solutions with the 2008 build agent. This article described a means to fix a problem with a similar error message and, even though it wasn't related to Sandcastle, I decided to give it a try.

Short answer: it worked!!

The fix feels more like a hacky workaround than a good solution, but absent anything better, and given the fact that it allows the build to actually function properly, it currently stands.

The solution is to change a setting in a .Net configuration file.
  1. Edit %Program Files%\Microsoft Visual Studio 9.0\Common7\IDE\PrivateAssemblies\TFSBuildService.exe.config.
  2. Find the line that assigns a value to the key MSBuildPath.
  3. Change the value from an empty string to the directory of the .Net runtime that should be used, e.g. %WINDOWS%\Microsoft.NET\ Framework\v4.0.30319.
  4. Save the file.
  5. Either stop/stop the TFS build agent service or reboot the machine.

I hope that helps someone else.

Tuesday, September 27, 2011

How to fix Visual Studio C++ Warning: LINK4075

At the beginning of last week's sprint, I picked up a task to reduce our warning count. One of the issues that I scratched my head over a couple minutes was a bunch of warnings with the text warning LNK4075: ignoring '/EDITANDCONTINUE' due to '/SomeOtherSetting' specification. As indicated by the LNK identifier, I knew the linker emitted this warning. However, in my perusal of the linker options, I couldn't find where the EDITANDCONTINUE setting existed. I even glanced through the other property pages looking for it.

Per my wont, I searched the web and eventually found it based on some oblique references on a StackOverflow question. The reason I had trouble finding it? It's a compiler option. And it's not called EDITANDCONTINUE.

This is controlled by the Debug Information Format setting on the C/C++ General page. If the option is Program Database and Edit and Continue, a.k.a. /ZI, then the output obj files apparently have a flag embedded in them that is incompatible with certain settings in the linker. Hence, the linker emits the warnings when it hits these conditions. It would have been nice for the message to have contained the /ZI option (which is what's used in the compiler) rather then than the EDITANDCONTINUE pseudo-option which isn't used anywhere, as near as I can tell.

To fix this warning, either change the linker options that are incompatible with the /ZI option or change the compiler option to something other than /ZI, such as /Zi. As of the time of this writing, the linker options incompatible with EDITANDCONTINUE include setting optimization to /OPT:REF or /OPT:ICF, turning off incremental compilation or setting one of /ORDER, /RELEASE or /FORCE. (These are from this MSDN page.)

Hope this helps someone.

Friday, September 23, 2011

Sharing folders with Windows 8 between VirtualBox machines

Currently Windows 8 runs just fine inside a VirtualBox machine. However, guest additions, the tools to enable certain useful features of VirutalBox, do not yet work with Windows 8. Today I wanted to use some Visual Studio projects from an existing Windows XP virtual machine in my new Windows 8 machine. The obvious first choice was to simply use a shared folder on the host in both virtual machines. However, this requires the guest additions, so I went looking for a work around.

What I found to work was a second virtual hard drive. In VirtualBox, on my primary development machine, I created a new hard drive on the existing IDE controller. I then started the machine and went into Windows' device manager. The disk manager started the new disk wizard. After initialization, I created and formatted an NTFS partition. I then copied the files I wanted to work with to this new disk. Finally I shutdown this machine.

Next, in VirturalBox, I added the same, new hard drive image as a second drive on the Window 8 box's SATA controller. (If you add the drive to the IDE controller, Windows 8 will try to boot to it first and fail with a bad image error.) When I booted Windows 8 and went into Explorer, the second drive showed up with all the data I'd copied to it.

That's it, pretty simple and straight forward, but not the obvious (to me) first solution.

Warning: I don't know what would happen if both virtual machines are run at the same time. I suspect (and hope) the second one started would get some sort of sharing violation and not allow the disk to be mounted. I can't imagine VirtualBox is smart enough to allow two machines to attach to the same vdi file at the same time without corrupting the data. However, it seems to work just fine as long as there is only one machine using the file at any given point of time.

Hope this helps someone.

Friday, May 6, 2011

Arduino notebook: 16x2 LCD display (Part 2)

How to change the default behavior

Prologue

My last article in this series talked about why I chose Sparkfun's Serial Enabled LCD display kit and how to build it. One of the reasons it caught my eye was because it uses an Arduino as the controller for the display. I figured it was pretty busy doing nothing most of the time and thought it could be used for other purposes in addition to driving the display. Upon further investigation, it turns out I was right.

Nicole, a friend of mine, recently e-mailed me about a photography project she was interested in doing but didn't want to spend the money for a professional rig to do some of the things involved. She asked if I had any ideas. I'd already given some thought to the problem, the investigation of which led me to the Arduino platform in the first place. This project was one I had stewing on the back burner for quite a while and it was now time to turn up the heat on it.

One of the requirements was to be stand-alone. It had to allow for some user configuration at the time of use in battery mode, without any other devices attached. This ruled out the obvious easy solution of setting the parameters via USB through a computer. Also, it involved controlling a stepper motor. I immediately thought of the LCD kit. I knew I would at least use it for the display, and hopefully some toggle switches. I had a distant hope I could also use it for the motor control. Since this was going to be a pretty permanent build, I immediately ordered another LCD to use specifically for this project.

After getting a new display on the way, I pulled out my first kit to take a look at it. Sure enough, it had a handful of empty holes on the PCB labeled with both digital and analog pin assignments. Cool. It had enough extra pins to accomodate everything I needed to do. So, I started work on the rest of the project while I waited for the LCD to arrive. (The other details of this project I'll write about in future articles here and on my photography blog Hooked On Light.)

First thing after getting the new kit, I assembled it. It went much faster this time with no of the frustrations I experienced on the first one. It actually worked the first time I applied power.

One of the biggest things I did differently was to use ribbon cable between the display and controller boards. In some ways this made things easier and in other ways it was harder. I think on the next project I'll probably put the header pins that come with the kit in one board and get a socket to solder onto the other board. That will give me a cleaner build like the first one, but with the ability to separate the boards for easier access like I have with the ribbon cable on the second one.

The web site to order the kit indicates the firmware can be reprogrammed with an FTDI Basic Breakout board. I had ordered one with the first kit, but hadn't had a chance to use it prior to this project. Knowing I needed to use this, I also purchased some 90 degree header pins from my local electronics parts supplier and soldered them into the holes on one end of the board designed for the purpose. This allowed the FTDI board to plug onto the controller board, providing a plug for a USB cable to be plugged in.

Now I started looking for instructions on how to actually reprogram the thing. The order page indicated you needed to change a setting in the Arduino IDE and referenced the github page for further instructions. Well, the github page had no other instructions on programming the thing. All it talked about was how to use the board through the serial interface. The software for the firmware was also hosted there. Besides the one sentence on the order page, I couldn't find any other instructions either on those two sites or anywhere else via web searches on how to reprogram this kit.

Reprograming the kit

I figured, no guts, no glory and decided to experiment some. I downloaded the source code from github, loaded it in the Arduino IDE and made a simple change to the splash screen text. Since the splash screen is the first thing you see, I thought this would be an easy way to verify if my changes worked or not. In the IDE, as instructed on the order page, I changed the target machine to Arduino Duemilanove or Nano w/ ATmega328 and hit the upload button.

It compiled.

It uploaded.

And when it reset after the upload, there was my new splash screen text.

Amazing. That was all there was to reprogram the kit's default software. The reason I couldn't find any other instructions was, there weren't any other instructions needed.

Epilogue

After this success, it was simply a matter of changing things to do what I needed it to do for this project. Essentially, I ripped out everything the existing software does and replaced it with what I needed. There are a few lines to initialize the LCD and one function to write characters to the screen that are the same, but everything else is completely new.

Now that I've found how easy this is to repurpose, I think I'll probably use it for other projects in the future too. It's an easy way to get an entire Arduino system up and running in a clean package with a simple user interface and minimal external support circuitry needed.

Monday, February 21, 2011

Arduino notebook: 16x2 LCD display (Part 1)

As I experiment with my Arduino kit, I find I want an easy, self-contained way to display information. The system provides one easy method in the Serial object with its print functions to send things to the computer through the USB port. This works well in development phase when the computer is attached to the board. However, I want to be able to run the kit detached from the computer and still get feedback. When there's limited data, I can use LEDs. They can be on or off to provide one bit of information. Additionally, they can be blinked in various patterns to communicate more than one bit of data. However, this is more abstract than I want. It doesn't strike me as easy, either to program or to use. I just want to read something directly without resorting to Morse Code to get the information.

I also don't want to spend much money.

So, with these three requirements in mind, I searched around. It seemed an LCD display was the way to go, as I found a couple different options. The first is to connect the LCD directly to the Arduino kit. At about $10, it is an inexpensive choice. It requires a number of wires and a moderate amount of programming to drive the display. Yes, this meets the cost and self-contained requirements, but isn't as easy as I would like. The second option is a kit with one of these LCDs and support circuitry to allow me to simply hook power, ground and a single data line to my Arduino. It uses the same Serial object to display information as the computer uses to receive data via USB. So, it satisfies the easy requirement much better, is more self-contained than the bare LCD, and at $25, still doesn't break the bank.

Upon further investigation, the support circuitry turns out to be an Arduino. I'm pretty sure this controller is sitting idle, spinning around doing nothing, most of the time. In addition to getting an LCD display, I believe this option in essence gives me a second Arduino that I might be able to reprogram to have additional features.

With all this going for it, I purchased this kit from SparkFun.com.

How to build a Sparkfun Serial Enabled LCD kit (LCD-10097)

Some assembly required

SparkFun Serial LCD kit - Step 1
(Click any of the pictures for a larger view.)

This is definitely a kit. It comes as seen above: a printed circuit board (PCB), some capacitors, a resistor, a crystal, a transistor, a potentiometer, a 28-pin chip, the LCD screen assembled on its own board and miscellaneous hardware. All this comes in a plastic bag in a box.

There were no instructions with the kit, nor could I find any online. To make up for this, the circuit board was well marked with silk screened values for the components and orientation marks for items that have to go in one way. Since I have put together many electronics kits over the years, I wasn't intimidated by this at all. If you are new to building kits, not having instructions might put you off a bit. However, don't let it. It's really pretty simple if you know how to solder. I'll outline what I did below to help you along.

As an overview, I basically installed the smallest components first and worked my way up to the larger ones.

SparkFun Serial LCD kit - Step 2
First, I installed the small, ceramic capacitors and the one resistor. There are two values for the capacitors. The capacitors labeled "220" go in the locations marked "22pF" on the PCB and "104" capacitors go in the locations marked "0.1uF". The resistor goes in the place marked for it. How these components are oriented on the board does not matter. Either lead can go in either hole.

SparkFun Serial LCD kit - Step 3
Next I added the transistor, crystal and electrolytic capacitor. The crystal can go in either way. The transistor and capacitor however need to be oriented correctly. For the transistor, make sure the flat side of the component matches the outline on the PCB. The capacitor has a light band with a minus sign in it running down one side. Make sure the lead next to this marking goes in the hole with the minus sign silk screened next to it on the PCB.

SparkFun Serial LCD kit - Step 4
Following this, I added the power/data connector and the potentiometer. These pretty much can only go in one way. Just match the components to the silk screen outlines on the PCB.

SparkFun Serial LCD kit - Step 5
After installing all the discrete components, I went to my local electronics store and picked up some DIP sockets. Generally, I don't like soldering IC chips onto boards directly. If I damage them with too much heat while soldering, or do something else stupid to cause them not to work, I don't want to go through the hassle of unsoldering all those connections to replace it. I've had to do that before and it's no fun.

The store didn't have any 28-pin DIP sockets in the narrow variety, but two 14-pin ones will sit nicely next to each other without any problems. On the PCB, note the silk screened outline. One end has a flat edge and the other end has a divot taken out. By convention, the edge with the divot marks the end where pin one goes. The sockets should also have one end that's flat with the other end notched. It's best to put them in with the same orientation as the silk screen. This makes things much less confusing. After the sockets are soldered in place, the chip can be inserted into it. The chip also has a notch in one end. Just line up all the notches and everything should be good.

SparkFun Serial LCD kit - Step 6
Finally, I did a good visual inspection of all my solder joints with a magnifying glass. The next step involves soldering the two boards together. Once this is done, it will be very hard to take them apart, so I wanted to be sure everything looked good.

There are three primary things to look for:
  1. Solder bridges. This is when solder spans two pads when it shouldn't. This is particularly easy to do on the pins for the IC sockets, but they can happen anywhere.
  2. Cold or missing solder joints. When working down a row of IC pins, it's easy to accidentally miss one. A cold joint is when one part of the joint, either the PCB pad or the component leg isn't hot enough to melt the solder but the other one is. Solder will flow onto one but not the other. Both cold joints and missing joints can cause intermittent or unreliable operation.
  3. Solder splatters. Sometimes solder will splatter or drop onto the board where it shouldn't be. Make sure there isn't any messy blobs anywhere. If you do find one, typically they'll just scratch off pretty easily.

After the visual inspection, I soldered the header to the LCD's PCB. Then I placed a small piece of cardboard between the two PCBs to function as a spacer. I placed the controller's PCB on the other side of the header's pins and made sure the two boards were parallel. I wrapped a bit of scrap wire around the assembly to hold them together while I soldered the two end pins. Once they were soldered, I took the wire off and finished soldering the rest of the pins to the controller board. When all the pins were done, I removed the cardboard and declared myself finished with the construction phase.

Batteries not included

Now that I had the physical assembly done, the next step was to hook it up to the Arduino. I went to the oracle of all things technical, Google. Unfortunately, with a moderate amount of searching, I didn't find any definitive instructions regarding what to do next. I found some forum postings that alluded to using the Serial object and some others that said to just hook it up, but didn't give any details.

I figured I probably couldn't hurt anything too badly with a bit of guess work. The connector for the LCD was clearly marked with 5V, Gnd and Rx, so I connected the red wire to the 5V pin, black wire to the Gnd pin and the Rx wire to the Tx pin on the Arduino. Typically in serial communication, transmit on one device goes to receive on the other. With the normal trepidation that I always have turning on a kit for the first time, I plugged in the USB cable.

The backlight came on and no smoke escaped. Well, that's a good first step. The SparkFun site points to some documentation on github, but it only talks about command sequences to control the LCD, it doesn't talk about how to talk to the LCD. It presupposes you know how to do that. I found a library that seemed to imply it worked with this kit, but didn't outright say it did. When I tried using some of the code from it, I got nothing. At this point I didn't know if it was a software problem or a hardware problem. I turned everything off, unconnected the LCD and took a magnifying glass to the kit. I looked at all the solder joints I could. Unfortunately, most were in the middle where I couldn't see. I examined all the legs on the chip going into the socket to make sure none of them got folded up underneath the package instead of going into the socket. They all looked good. I hooked everything back up and tried again. Nothing. I tried more searching online to see if I could find more information. Nothing. Arg. Frustrated, I set it aside for a week.

When I came back to it, I rechecked all the hardware I could with the magnifying glass. I plugged it in and tried writing to it. Same result as before. The backlight came on but I couldn't get it to display anything. Looking at the library code, it seemed a bit suspect, but even if it was wrong, I thought I should still see something. I searched more online but found no new information. One last time I decided to take a look at the kit before contacting customer support. And then it hit me. Pin one was in the wrong position. I couldn't believe it, I'd stupidly put the chip in backwards. I pulled the chip out and put it in the correct way. This time when I applied power, it showed a splash screen. Oh, much, much better.

I uploaded a program that I expected to put some data on the screen and it worked. Well, sort of. That code that looked suspect in the library I used didn't work properly. I suspect it was for some other serial controller with a different command set. I threw together some code based on the manufacturer's documentation on github and suddenly it worked properly. Woo, hoo!

Lessons learned

  1. If you build this kit, expect to see a splash screen on power up. This will tell you if the board works or not, independent of serial communication.
  2. Make sure to get some sockets for the chip. I'm sure glad I did. It would have been a nightmare if I'd accidentally soldered the chip in backwards and then had to desolder first the header and then the chip itself to correct the mistake.
  3. If I ever build another one, I will get a plug for the header. This would allow the two boards to plug together rather than be soldered together. A ribbon cable could also be placed between them for more flexibility with mounting.

That's about it for this write up. Next for this project is to figure out the best way to communicate with it. I need to look at the library to see if I was using it wrong or if it really is designed for a different unit. After that, I plan to investigate reprogramming the LCD's controller to investigate using it for additional processing tasks.

Tuesday, September 7, 2010

Synergy setup between a MacBook and Linux machine

I just got a new MacBook Pro a couple days ago. Wow, it's a nice machine. For my first Apple product, I'm pretty impressed. I have a couple things I don't like about it that I'll probably rant about some other time. But for today's post, I wanted to chronicle my adventure in getting it to be a Synergy server for an old Dell Dimension running Ubuntu. All said, it was the most problematic software configuration I've had in a really long time, but in the end I got it working pretty well. I kind of reminded me of trying to sort out IRQ and DMA addresses way back in the dark ages.

I first heard about Synergy a couple years ago when a friend of mine connected his iMac and Windows boxes together. For those who don't know, Synergy is an open source project that allows one machine's keyboard and mouse to be used on another machine in a seamless manner. The machine that has the keyboard and mouse physically connected is the server and the machine that only has a screen on it and uses the input devices from the other is the client. The two machines are connected through the network. So, this solution doesn't require any additional hardware than what the computers probably already have. Once configured, the mouse pointer can be moved off the side of one machine's screen onto another and the keyboard focus follows. It can also work with more than two machines. Any number of boxes can be setup as clients to use the common server.

Official installation instructions can be found on the Synergy wiki. I followed them mostly. Below are some issues I ran into and how I worked around them.

Step one was to download the software from the repository. For the Mac it came as a dmg file that installed without issue. For the Linux machine, it was available in both deb and rpm formats. Since I'm running Ubuntu, I loaded the deb file. It installed just fine too. In both cases, it put two applications in /usr/bin: the client, syngergyc, and the server, synergys.

Step two was to configure the server. This is done through a text file. The instructions seem to imply that it is optional, but it isn't. This is the only way to configure the server. Another point not clear in the instructions is the configuration file is only needed on the server and can be anywhere and called anything. You specify the name, including the path, on the command line for the server.

Somewhere I heard about GUI configuration tools; I thought it was on the setup page, but when I went to find it, I couldn't. Where ever I heard about them, I tried SynergyKM on the Mac. It was a dismal failure. It seemed to get confused and couldn't maintain its state very well so I eventually abandoned it, but not after wasting some time trying to get it to work. In the end, I just wrote the configuration file myself. It wasn't as hard as it first seemed. I called it /etc/synergy.

I also found a GUI tool for the Linux client and tried using it. However it used a different version of Synergy and when the Ubuntu Software Center tried to install it, it somehow messed up the apt state files. Another weird problem. I haven't had apt files get messed up in quite a while. I had to go into aptitude and do some clean-up to repair things so the GUI tools would work properly again. In the clean-up I also removed the GUI tool and just handled setup manually in a terminal window.

After installation and configuration, the next step is to run both the server and client in a terminal to test connectivity. My suggestion is to not skip this step. I thought I could and ran into some opaque problems that became obvious when I ran the software in a terminal. They all related to machine names used and my getting confused about which names to use where. By default, the instructions say the hostname is used, but that didn't really work too well for me on the Mac, so I ended up specifying my own names for the synergy connections. It took a bit for me to get this to work on both sides, mainly due to my own wrong assumptions about how things worked. When I ran things in a terminal, the error messages indicating my mis-configuration made short work of getting things sorted.

So, I once things ran in two terminal windows, the next step was to start the software automatically at startup. On the old, deprecated site, there's an Autostart Guide and this is linked to from the code.google.com wiki setup page with some additional information. I didn't see either on the new, official site and had to do some experimentation using the information from these two pages to get everything working properly. I ran into several time consuming problems doing this.

The first problem was a how to connect to the MacBook by name. The Synergy client needs either the IP address or the DNS name of the server in its command line. This makes sense from a client programming perspective. However, from a system administration perspective not so much. I'm using DHCP to assign IP addresses and as I have a small home network, I don't have a DNS server setup to handle local requests like this. The standard name resolution mechanism didn't work for me. I wanted to use the netbios name since it was available but it took a bit to figure out how to get it. In the end, I found the nmblookup command which, given the name will give some information, including the IP address. But, it's not in an incredibly useful format for scripts. I had pipe the output to grep to get just the line I wanted and then pipe that to cut to get just the IP address. Specifically the command looks like this:
nmblookup macbook_name | grep "<" | cut -f 1 -d" ".

With this in hand, I wrote a short bash script to look for a server and then connect to it once it found one. This is where I ran into my second problem. It's been a while since I've done any shell programming and I had to do a bunch of web searching to remind myself of the details. Everything indicated string equality could be done with either "==" or "=". As a C programmer, I defaulted to using "==". My script kept giving me unexpected operator errors on odd lines without operators. After commenting some lines out, I found the if statement at fault. On a whim, I changed the operator to be "=" and the problem went away.

The final version of the script became this:
#!/bin/sh
svr_ip=""
until [ "$svr_ip" != "" ]
do
svr_ip=`nmblookup macbook_name | grep "<" | cut -f 1 -d" "`
if [ "$svr_ip" = "" ]; then
sleep 2
fi
done
synergyc $svr_ip

(Note: right now this will fail if the server reboots and gets a different IP address since once syngeryc starts, it continues running even when the server goes offline.)

Once the script was done, I then followed the instructions to modify the gdm files, substituting my shell script for the synergyc call.

When this was all working, my next step was to get the autostart working on the Mac. By and large, this went faster than the Linux side but not without its own problems.

The first one was that the install location of the synergy software was different from the examples. So, the first time I tried it, it simply didn't do anything. There were no errors anywhere that I could find pointing to the problem. It was a matter of going through the script line by line until I found it.

With the software's path corrected, I ran into the second problem. When I logged on, the screen went blank and I never got anything else. The menu bar never appeared. The dock never appeared. The desktop icons never appeared. I couldn't do anything. However, the mouse moved the cursor and it would move onto the Linux desktop. Ok, I knew it was running, but how to get my desktop back? As a new Mac user I had no clue.

After a bit of searching, I found I could boot the installation CD and fire up Terminal to get to my system. I inserted the CD, held down the power button to turn the machine off and then held down the "C" key as I turned it back on. As advertised, it booted the CD and I found the terminal in the Utilities menu. I edited the script and added an ampersand to the end of the call to start the server so it would fork the process and run in a different thread. After rebooting, everything finally worked as expected.

As I said up front, once I got it working, it works really well in spite of the problematic configuration. The only issues I've found are relatively minor: 1) copy/paste doesn't work between machines, 2) the screen saver on the server doesn't cause the screen saver on the client to start, 3) I haven't gotten it to work prior to logging onto the Mac and 4) the above noted problem if the server reboots and gets a different IP address. While it'd be nice if these worked, in the greater scheme of things, they are pretty minor failures and I'll live with them for now. I think with more time the last two can be overcome with simply configuration changes that I haven't figured out yet.

Thursday, December 10, 2009

How to install Jolicloud in a VirtualBox

I ran across Jolicloud today as I surfed around the internet. It sounded like a competitor to Google's Chromium operating system and sounded interesting. Just for investigation purposes, I wanted to install it in a VirtualBox virtual machine so went to Jolicloud's download page. It gave a page of instructions that indicated I needed to download an operating system specific application and an ISO file. The application would use the ISO file build a bootable thumb drive. This drive could then install Jolicloud on the target notebook.

Not wanting to go through this hassle for a virtual machine, nor knowing if I could even boot a virtual machine with a thumb drive, I searched for instructions for a non-thumb drive install. I found several sites but apparently, they were for an earlier build that had some other bootstrap process. They mentioned using an .img file (which the current download doesn't have) and going through a number of different steps to convert it to something the VirtualBox software could read. These directions didn't seem to be appropriate for the current version. (These sites indicated they were for an Alpha release that required an invitation. At the time of this writing, the version is a PreBeta and does not require an invitation.)

I let the problem percolate in the back of my brain for a bit as I worked on other things. At some point, I realized the current build uses an ISO file. I wondered, "What are the odds that image is bootable?" Figuring the odds were good, I downloaded all 600 megabytes of it.

When the download finished, I used VirtualBox's normal creation wizard to create a simple virtual machine (512MB memory, 4GB virtual hard drive, Ethernet bridged to the host's NIC). I mounted the ISO image as the CD drive and set it as the boot drive with the virtual, empty hard drive second. When I started the machine, it booted just fine and presented a standard startup screen giving me several options, one of which was to install to the hard drive. From this point on, it proceeded as a typical install. The only things the install wanted from me were the language, keyboard layout, time zone and user information. When done, it shutdown. I unmounted the ISO file from the CD drive and turned the machine back on. Everything seems to be working perfectly so far. For whatever it's worth, it does seem to be faster than Chromium running in another, identically configured virtual machine.

That's it. Very painless. Everything just worked.

Friday, June 26, 2009

How to display tool tips in WPF DataGrid text column

Like many people before me, while using the DataGrid from the WPFtoolkit, I recently had a requirement to display text nicer than the default. First, the default is simply to truncate the text like so:

I needed the truncated text to have ellipses after it and have a ToolTip showing the entire text. Something like this:

I did a lot of searching the web and didn't find anything straight-forward and elegant. I found things that intimated using styles in XAML might work, although nothing directly addressing what I wanted to do. Furthermore, I found others saying XAML styles won't work at all for cell level changes along with some ugly hacks where, inside the LoadingRow method, they walked back up the VisualTree to change the cell style in code. My search was confounded by the fact that there are a number of different DataGrid components, three by Microsoft (WinForms, WPF and Silverlight) and several by other 3rd party vendors. All have issues doing this very basic thing and all with different workarounds.

After spending the better part of a day researching and testing various possible solutions, I found a comment on a Silverlight forum with the idea of using a DataGridTemplateColumn descendant to handle the new needs. I started to adapt the example code from Silverlight to WPF and thought, "There must be an easier way." Since the existing DataGridTextColumn already did much of what was in this example, I went off to look at its source code. While browsing through it, I found a protected virtual method called GenerateElement that apparently is a factory for the cells' display. On a hunch, I created a descendant of this column type and overrode this one method, changing the properties of the returned value as needed.

Amazingly, it worked the first time. So, to add my own workaround to the mix that's found on the web, here is the class I created:
public class DataGridToolTipTextColumn : DataGridTextColumn
{
    protected override FrameworkElement GenerateElement(DataGridCell cell, object dataItem)
    {
        var result = (TextBlock)base.GenerateElement(cell, dataItem);
        result.TextTrimming = TextTrimming.CharacterEllipsis;
        ApplyBinding(result, FrameworkElement.ToolTipProperty);
        return result;
    }

    // Copied from DataGridTextColumn because it's not protected there either. Seems like it should be.
    internal void ApplyBinding(DependencyObject target, DependencyProperty property)
    {
        var binding = Binding;
        if (binding == null)
            BindingOperations.ClearBinding(target, property);
        else
            BindingOperations.SetBinding(target, property, binding);
    }
}
The GenerateElement method simply calls the base class and typecasts the return value to a TextBlock. For the moment this is safe since the base class has the type hard coded. The TextTrimming property is then set for CharacterEllipsis and the ToolTip property is bound to the same value as the Text property. This class is twice as long as it needs to be because the ApplyBinding method that exists in the base class is marked internal rather than protected, so I simply made a copy of it for use here.

In my usage, this DataGrid is only to display information; editing is turned off. I haven't checked to see if this would cause any issues in edit mode.

Related material

In all this searching, I did find a couple good resources:
  • Samuel Moura has a 4 part series doing some pretty intensive DataGrid styling and templating: Introduction, Custom Styling, Playing with Columns and Cells and TemplateColumns and Row Grouping.
  • The Tranxition Developer Blog has a good article on adding dependency properties to the TextBlock type. Unfortunately, this doesn't seem to work for TextBlocks that are in another control's VisualTree.
  • This Silverlight thread is what gave me the idea for creating a descendant class.
  • I ran across this great little utility called Snoop to inspect the VisualTree of a running WPF application. I think I'd heard of it before in passing but had never looked into it. I don't know how many times I've written throw-away VisualTree walkers to dump information to the console. Now I'll never have to do that again.