One of the things about being on Channel9 and other similar sites is that I always have to know what other developers think before I embark on a big project. Also, since I picked up C# and .NET, something about the environment always has me pushing to find out what the best practices are before beginning on my own with something I haven’t done before.

Weird. I didn’t used to be like that. I usually solved these problems completely on my own, but now I’m always paranoid that I may not be doing it the best way.

Anyhow, a quick question..

Exporting data from a Web application to be used on other Web sites or desktop and mobile applications is a no-brainer for me, but what about the other way around? What if I have an application that must share data with an ASP.NET application?

I’m trying to give a desktop server application a Web front-end to allow users to check status remotely, configure it, and pull data from it. My data can be easily represented in simple XML in this case. SOAP is not important. I don’t necessarily need to pass objects back and forth, just data.

What are my best options? I’m using C# for the desktop server application and C#/ASP.NET for the Web application, although it would be nice to not be limited to just .NET on the Web application side.

Remoting, named pipes, and proprietary UDP and TCP solutions all came to mind. I’m thinking since objects aren’t important and it would be cool to not limit myself to .NET on both sides, remoting might be out. And I might want a “session-less” solution for efficiency since I’d be fetching data between Web page loads, which rules out TCP.

But, anyway, does anyone have any ideas or has implemented something like this before?

Thanks in advance!

AT wrote:

P.S> It will be good if you remove all your threads and use also async IO for file I/O. In case of completely Async processing you will get maximum CPU performance.

I have a multithreaded TCP server that must rely on synchronous communication (blocking I/O). Each connected client to the server runs in its own separate thread. Each thread must be able to communicate with each other by sending “messages” to one another. Additionally, some events must occur at a timed interval (such as, if a client hasn’t authenticated itself in three minutes, it will be shown an error message and be disconnected.)

How I solved this in Delphi is to have the blocking read operation on each socket time out every 500 milliseconds and that’s when I’d check to see if any scheduled events were supposed to take place. I’d also check the client thread’s queue to see if any other threads left “messages” for that client and then process them. Then I’d block for 500 more milliseconds until some data arrives, and so on.

When porting this application to .NET, I am having some trouble because all of the socket blocking is done at a lower level. I can’t use the blocking operation iterations of the socket to check my queue or check for scheduled events. I’d much rather separate socket operations from my application logic anyway, but I need a strategy to allow me to both block socket operations and check my queue and timed events.

How do I go about this one? I could set the socket to have a receive timeout every 500ms, but then an exception is created if the socket times out. But I’m not timing out the socket to disconnect it, simply to interrupt it for a while so I can check my queue and do other stuff. So I’d loop again. That means potentially countless exceptions are created and caught per thread, and that’s not good for performance.

I tried peeking the socket, but that causes a noticeable delay to the client and it becomes difficult to tell when the socket has suddenly disconnected. This is due to the fact that I must call a Receive() eventually to tell if the socket’s still there, and that will block indefinitely.

What are my options? I don’t even really want to have a queue of messages at all, I’d much rather have them event driven this time so I don’t have to poll to check for messages. But this is complicated in a multithreaded application because events are handled by the same thread that triggered them. That particular problem was easy to solve in a GUI application with Control.Invoke(), but this is not a GUI application.

Any ideas?

lenn wrote:

I love this thread.  Although I am not sure what the answer is yet, part of me wants to just stand back and watch you guys do something cool together.  Before I know it there will be a project on source forge.  Part of me wants to start a community project to build a new piece of Channel 9. 

Let's talk about this for a sec.  What is Channel 9 lacking that you would just die for but we may not get to fast enough with our little two man dev team and our day jobs? 

As far as coding contests go ... stay tuned

gmiley wrote:

Another quick note. GPL doesnt mean companies cannot use the software, it just means they cannot sell it or incorporate it into a product they intend to sell.

By GPLing the projects of the contest it would ensure that anyone that wanted to use it could use it. Anyone that wanted to learn from its source would also be free to learn from it. If the end product is something extremely nice, by all means emulate it and make some cash by selling it, but I'de prefer to keep the originals in the public domain.

gmiley wrote:

To do this though, we would need a designated, unbiased team of professionals to judge who wins?

All restrictions must be known for each project: time limit, language, compiled project size/lines of code, etc...

A merger of the two ideas would be awsome. People submit ideas they can't afford to do, a contest is held to produce the application and the winners code is licensed under the GPL or LGPL, the LGPL is a little more flexable, but not sure how everyone would feel about it. Remember just because this is a MS site, it doesn't mean "no Free software". Besides I think a community project should be exactly that. If we, as a group, write an application, it should be free for all to use and learn from.

We could compete to accomplish the best possible solution to a particular problem, backed by design and coding. The competitions could even be for things used by the Channel 9 site, something practical.

We could lock off team forums so that only members of the team can see current team discussions and it’d be a secret to how we solved our problem until the end where we’d all present our solutions. Then there could be some sort of public voting system that would determine the winner.

I don’t know, just throwing some ideas out there. I love the idea of pitting the Five Guys against each other and even working for one of them.

Sure it would introduce a gigantic mess of chaos, but if we could put together a system that would effectively delegate responsibility, host helpful design meetings, and control source code and contributions in a community this big, then we would be building a collaborative system that would rival most in the world.

Channel 9 has been a gigantic leap for Microsoft in my opinion. I mean, it’s been such a conservative company in the past. I think an idea like this is the next big step and we’d learn so much from each other.

Why would video be a problem? People could submit their own. Some of us have only worked in one or two different industries and some of us (like me) have only just graduated from college. I’d love to see how people from different companies of different sizes in different countries tackle the same problem.

Services are really popular now for firewalls, antivirus software, database engines, etc. My current service application that I’m working on is conference software that allows users to chat with one another through multiple interfaces (telnet, specialized GUI client software, and the Web). Running as a service has a lot of advantages: (1) it can be controlled via the service controller locally or remotely from another computer, (2) it has built in failsafe if the server should go down (you can give services instructions on whether to restart or reboot the machine or whatever), (3) it can run when no user is logged onto the machine and users logging on and off don’t disrupt the server at all.

My service application has multiple services in it (which is one reason a tabbed view might be cool, and maybe a tree node view for service applications so you can see all the services that belong to them them), such as the interface for telnet, the interface for smart clients, and the interface that exports Web services. It also has an NNTP service for newsgroups. I’m not the worst offender though. Oracle has like 10+ services.

With all the attention services are receiving today, I thought maybe the service controller would be organized more like the Global Assembly Cache in .NET which is pretty well guarded against naming and versioning conflicts with strong naming.

As for SQL Server, Exchange, or BizTalk, I’m not sure. They’re not core services and they’re not third party either. Maybe you should be able to create different sections like you can in the event log.

I have noticed that Windows services are becoming increasingly popular. I have created a few of them myself.

What I’d really like to see is some sort of logical and physical division in the service controller that divides the core operating system services from third party services. For example, in the management console there could be some sort of tabbed view that divides the core system services from third party services. Additionally, there could also be some sort of visual cue that gives attention to the critical services that the operating system absolutely must rely on and will not shutdown.

I also use “sc” from the command line to control services so it’d be great to see some sort of division there as well.

A physical division between the two types of services would also be nice so that third party services could be completely wiped out and the system would be unaffected. (Maybe this is too optimistic of an idea because it could create some problems with third party services that provide low-level hardware support.)

I’ve also noticed that we’ve come a long way with version control and resolving naming conflicts, yet these ideas are almost absent in the service controller. Since most services sport generic names, this could only lead to trouble as more and more people move to NT platforms and services get more attention.

1)      One thing that really gets to me is when Explorer windows steal focus from other applications and sometimes even themselves. Sometimes after a transaction (such as Internet Explorer downloading a Web page) or worker threads updating information, these windows will steal focus from other applications as if you had pressed ALT+Tab. Sometimes when I launch IE and type in an address in the URL edit box, IE will reset focus to the edit box as I’m typing. That means when I am typing in “www.google.com” my typing will come out to be something like, “.com www.goo  WTF?” because the edit control’s focus was reset and the caret moved to the beginning of the line.

2)      I work extensively in console windows because I find it a lot faster than clicking around or even using keyboard shortcuts in Explorer windows. Just about any kind of event on the desktop or within an application such as a popup window or instant message or a call to the API for FlashWindow will take me out of a full screen console window. FlashWindow is the biggest offender, taking me out of full screen sessions after each and every instant message was received from just about any instant messenger.

Although, the new “Filter by” feature in Longhorn is extremely handy and I love whoever thought to finally implement it in Explorer windows. I find myself using the console window a lot less now because of that feature.

3)      After resizing a console window to the perfect size, as soon as I type in the window or some text comes in by itself triggered by some event that puts output on the screen, the console window will completely resize itself and even select a different font. No one told it to do that!

4)      I get locked shares where Explorer tells me a file is in use by another process even if I close all shares from Computer Management or come back fresh from a reboot. These are not system files and are definitely not in use, and require me to delete them from console recovery mode or from another operating system (thank god I have Longhorn on dual boot.)

That’s it for now. This is one of those threads where I’ve been dying to say something for a while, but then when you’re on the spot you have trouble thinking of something.