Motivation
I recently decided to cut the cord on cable and downloaded
PlayOn to do the job of serving video content from the internet to my Xbox 360 and Wii, so I could watch it on my TV. Naturally, this led me to figure out how it worked, and how I could play around with it.
One thing I immediately wanted to do was improve the organization functionality. I thought I'd start with a plugin that simply takes the media server's location of video output and provides it as input back to the same server, but in a different organization scheme. Unfortunately, that didn't work as PlayOn didn't know what to do with its own URIs as input for a
VideoResource in the Lua plugin API.
I didn't want to get into the scraping, transcoding, and serving of files from the internet -- that's already done well and I didn't want to spend all that time creating a media server of my own. But I did want to see better fastforward and rewind capability. To solve that, I thought I'd create a
DVR for PlayOn (or TVersity, or any other media server really) and knock out both the organization features I wanted, along with the seek functionality.
Launching a business
This will be my first attempt at launching a money-making venture online (aside from programming for other people's money-making ventures). I don't expect this will turn into a full-time job, nor do I expect I'll make retirement money off of it, but I think it can make a decent return that won't require a ton of ongoing work on my part, and it might make a fun experiment.
Just as well, I thought it could make for some interesting blogging for me, and reading for you. Here's how I plan to begin:
I'm starting with a simple landing page that asks for your email address, telling potential customers we'll let them know when it's ready. Cost: $22.00 or so for the .net and .com domains. Email list is free to start out at
MailChimp, and I'll only need to pay them if I generate more than 500 signups and want to email those folks (which of course, I will).
The reason behind the landing page and email list is to gauge interest before I start. Right now I've
only tested the technological feasibility of my idea by trying out all the moving parts needed to make it work. It's not all in one code base (some of it isn't even code), and there's no UI to speak of.
If there's a decent amount of interest, then I'll put it all together with an easy to use interface, and I'll probably devote more time to it in the short run. If it looks like me and my tech savvy friends will be the only users, I'll probably still release it, but you might configure it with text files instead of WinForms until I get sick of doing that and build the UI for it later down the line. I'm pretty sure that would be a free version. =)
So by going the landing page + email list route, I only invest a little time and money up front, until I can see what the demand will be like. I seem to remember hearing that
dropbox started similarly, and I'm sure I must have heard this tactic mentioned a hundred times on Mixergy, where Andrew Warner
interviews entrepreneurs about how they've been successful (and sometimes not-so).
On to pricing: I'm thinking of initially starting it out at a one-time fee of $35 (US), but that may change. More than likely, I'll test different pricing schemes and see what works well. I've got $105 of credit in Google Adwords, and I thought I might use that to do some A/B testing on the signup page. Half of the users would see Test which A might mention the one-time $35 price point, while the other half see Test B, which may be $5 per month. Hopefully, this would inform us as to which one results in more signups to the email list.
However, I decided not to do that at this point. It's just more up front work that has to be put into the idea. Instead, I'll probably wait to see if anyone signs up for the list, then do something similar to test which pricing options are better than others (using the email list instead of the website).
As I mentioned above, the DVR is not ready for public consumption yet, but you can sign up for the email list on
MediaServerX if you want to be notified when it's ready.
If this little business experiment generates any action on my part aside from "well that flopped, I give up," I'll keep you updated here on the blog.
Deciding on technology
At first I thought I might like to really dig deep and write something from scratch, so the world was open for me to be cross platform and there were flowers in the fields with
unicorns and rainbows.
However, after taking a look at the
UPnP spec documents, I realized I didn't really care about that rainbow and what I really wanted was to just write some code.
So I moved on, thinking I'd use
Developer Tools for UpNP Technologies to generate the beginnings of a
control point. After messing around with it for a little while and reading that Casey Chesnut "
would probably try to use the MS stack next time," I decided to go with the
UPnP Control Point API that's been included on Windows since Millenium Edition was all the flop.
It seemed like the lowest barrier to entry, which is something I haven't heard in a while about programming and Microsoft.
Code
First I created a solution in Visual Studio and included a reference for the UPnP 1.0 Type Library (Control Point). It can be found by right-clicking on References, then clicking on Add Reference. It's under the COM tab.
To start out doing anything, you can find a UPnP device using the UPnPDeviceFinder class. The code below will find all devices listed as MediaServer:1 (there are also :2 and :3, but I have yet to look into the differences between them).
UPnPDeviceFinder deviceFinder = new UPnPDeviceFinder();
UPnPDevices mediaServerDevices = deviceFinder.FindByType("urn:schemas-upnp-org:device:MediaServer:1", 0);
foreach (UPnPDevice mediaServerDevice in mediaServerDevices)
{
// do something with the UPnPDevice
}
The basic object model consists of a Device Finder (seen above) which can get you a collection of individual devices. Individual devices have a collection of services. A service is what you need to communicate with to control and get information from the device. (
Here is a diagram).
In our case, we want to get the MediaServer's ContentDirectory service to give us a list of ContentDirectories and their children.
UPnPService cdService = mediaServerDevice.Services["urn:upnp-org:serviceId:ContentDirectory"];
Now that you have the content directrory service, you can invoke actions on it. For example, to see what it contains you can use the browse service:
string browseFlag = "BrowseDirectChildren"; // BrowseDirectChildren or BrowseMetadata as allowed values
string filter = "";
int startingIndex = 0;
int requestedCount = 1000;
string sortCriteria = "";
object[] inArgs = new object[6];
inArgs[0] = objectID; // use 0 for root
inArgs[1] = browseFlag;
inArgs[2] = filter;
inArgs[3] = startingIndex;
inArgs[4] = requestedCount;
inArgs[5] = sortCriteria;
object outArgs = new object[4];
cdService.InvokeAction("Browse", inArgs, ref outArgs);
object[] resultobj = (object[])outArgs;
string result;
int numberReturned;
int totalMatches;
int updateID;
result = (string) resultobj[0];
numberReturned = (int)(UInt32) resultobj[1];
totalMatches = (int)(UInt32) resultobj[2];
updateID = (int)(UInt32) resultobj[3];
This where it really starts to get wieldy. I find it a complete PITA to be sending and receiving basic
objects and casting them back and forth to their desired types. The
result string ensures more fun searching through XML to get what you want. It's nothing we want to be doing throughout our source code, because it's a ton of noise to sort through.
Side note: Setting the objectID to 0 will get you the root element. After that, the results report back their IDs so you can use what it returns. That took me forever to find out, so hopefully it will save you some time.
Obviously, this sort of hassle calls for some abstraction, so I began writing a wrapper library. So far it works through everything I've shown above, and includes a sample project showing how to use it.
Source code:
upnplib-mediaserver-wrapper
It's a Visual Studio 2008 solution with 2 projects: the upnplib-mediaserver-wrapper is the library, and example-upnp-mediaserver-browser is an example of how to use the library. You'll need to set that one up as your startup project for it to run. I'd have guessed it would be in the .sln file but it's not, and I don't want to include my user-specific settings in the source control.
Let me know if you're interested in seeing it become more complete.
If you're a fan of
PlayOn or TVersity and want a DVR service, click the preceding link to sign up and be notified when it's ready.
As always, I'd love to hear your thoughts or questions on the subject.
Hey! Why don't you make your life easier and subscribe to the full post
or short blurb RSS feed? I'm so confident you'll love my smelly pasta plate
wisdom that I'm offering a no-strings-attached, lifetime money back guarantee!
