|
|
Comments and Discussions
|
|
 |

|
Hi, I have read the article you wrote, and I need some help here.
Here's the problem: I need to take a
file, the file may be any file, opened with something like a Browse button, and
to encrypt it, then send it to a specific address, the addres on the net is
also given by me, and there, if the file arrives safe at the address I wanted,
it will be decrypted and can be read by the receiver. I have some code, but I
have to use the sockets programming stuff, and to know the specific port also,
so that , if I would like, the program may run on the same machine as well,
just sending and receiving at the same port.
Can you send me someinfo and code about that?
Thanks,
steve
|
|
|
|

|
But DCOM wasn't perfect; it introduced new complexities. Like COM, whenever a server-side component is updated using DCOM, the type library information changes due to binary incompatibility. With DCOM, these changes need to be propagated to existing client machines. DCOM doesn't provide a mechanism for dynamically updating and binding to type library information; such information is stored in the registry, or with COM+ in the COM+ catalog. DCOM is a "chatty" protocol, pinging clients regularly to see if the clients are still alive. And because it doesn't support batch operations, it takes almost a dozen roundtrips to the remote server to complete a single method call. Using DCOM through firewalls becomes problematic because it dynamically allocates one port per process (configurable through the registry) and requires UPD and TCP ports 135-139 to be open. An alternative for enabling DCOM through firewalls exists by defining Tunneling TCP/IP as the underlying transport protocol. This allows DCOM to operate through some firewalls via port 80. But it's not very reliable, doesn't work through all firewalls, and introduces other limitations (lack of callback support, etc.). DCOM has certainly evolved over the years, in an effort to accommodate the demands of a changing environment. But because of its roots in older binary and component-based protocols, it still fails to deliver the flexibility needed in today's enterprise. DCOM is still inefficient, cumbersome to deploy and requires a fair amount of manual maintenance. From the site http://msdn.microsoft.com/library/enus/dndotnet/html/webservicesdcom.asp The author points out the problems of DCOM, but does not give out reasons clearly. I think the most fundamental problem of DCOM uses the net communication pattern 1! Yuancai (Charlie) Ye See 30 OLEDB samples Use of free SocketPro for creating super client and server application www.udaparts.com
|
|
|
|

|
Hi, All
I updated SocketPro with its documentation and more samples. See the site http://www.udaparts.com. It will takes a while for Codeproject to publish my this updated article
Yuancai (Charlie) Ye
|
|
|
|

|
Very excellent, but can You create a demo for transfer huge from server to client ?
Thanks.
Ivan
|
|
|
|

|
Contrary to other readers, I appreciated your article.
I worked myself on a similar project and I am curious to see your implementation.
Have you heard about GMSOCKS?
Regards
Jeff
|
|
|
|

|
where is the document of your com library.
|
|
|
|

|
First I really get a feeling you don't understand what DCOM/CORBA does.
DCOM/CORBA is sooooooo much more than just plain socket transfer.
1)Marshaling Data? Oh, You expecting IP address? Here we go have fun with my Picture scanned at 14000x14000 dpi.
2)Do you handle security in SockPro?
3)Do you release Interfaces at sometime???
MTU/Nagle etc.
Nagle algorithm is turned on by default. What make you think CORBA and DCOM turn it off?
>However, these technologies have one common and fundamental problem that all
> of calls between a client and a server are blocked for a returned result
> (please correct me if this is wrong). A client stays still, and has to wait
> for a while after sending a request to a server. If the request is a
> lengthy action, the client application seems to be dead to a user.
In other words you want to be asyncronously notified when action is complete?
Hmm, Chapter2 Event sinks/ConnectionPoints. Create threadpool, assign worker.. notify back.
So let's see:
>All of applications developed from this framework will has the following BIG
> BIG BIG .... advantages plus other numerous advantages:
> 1)Non-blocking/blocking, parallel computing client and server.
Same as others.
> 2)Both client and server run at the fastest speed because of Nagle algorithm.
Nope could not be that. Less overhead because you hardcoded security + object, dropped wrappers - yes.
> 3)Cross-platform data communication and cross-language development
Use CORBA.
>ware, SocketPro still runs 30% faster even though SocketPro runs in the
> blocking mode according to my experiments.
Well this could be, but if you are, like me, storing data in the database - database runs 10x slower than anything else. Is file transfer good way to test DCOM performance?
>Other advantages includes:
>Simple installation and security setup.
Just about only place I'll give you some credit. Been long time but I still remember the bruises of DCOMConfig.
|
|
|
|

|
Sorry, I didnt see this post till it was too late, so I have enetered a message on the first one
mea culpa, I'd overlaid the browser session with another window, so only saw the top 1/2 of the index screen !!
Garth
|
|
|
|
 |
|
|
General News Suggestion Question Bug Answer Joke Rant Admin
|
A set of socket libraries for writing distributed computing applications over the internet
| Type | Article |
| Licence | |
| First Posted | 2 Jan 2002 |
| Views | 130,428 |
| Bookmarked | 59 times |
|
|