Hi there,
I have a Win32 graphics manipulation program, in which the manipulation code is
cleanly separated off from the GUI code in a statically-linked library.
The application loads and displays images, then combines them together in
financially pleasing ways using the library, then displays the new images back
in the GUI.
We now want to "split" this application in two, so that the GUI resides on
Client PC1, and the library runs on Server PC2.
While several communications protocols are possible, it would seem that
socket-based communication would be preferred means, using either Borland's
TTCPClient and TTCPServer, or Indy's TidTCPClient and TidTCPServer. Is this so ?
Also, the demands on memory on the Server are extreme, so we only want to work
on one "job" at a time - the opposite of the usual case. Two simultaneous jobs
could cause both to fail.
Finally, the communication needs to be a bit of a conversation, which goes like:
Client PC1: Hello - I'm PC1 - here's my username and password.
Server PC2: (server check credentials) - Hello PC1 - you're in - what do you want ?
Client PC1: I want to combine these two images using method A.
Server PC2: (server checks that this client is allowed to do this) - OK - send
me image 1.
Client PC1: Sends Image 1.
Server PC2: OK - got Image 1 - send me Image 2.
Client PC1: Sends Image 2.
Server PC2: OK - (creates Image 3 = 1+2) - and here is your combined image -
Sends Image 3.
Client PC1: Thank you - good bye.
Server PC2: (writes transaction to log; closes connection; can now accept other
connections)
So my final questions are:
1. which is better for my needs - the Borland or the Indy components ?
2. can anyone point me to some good tutorials / code examples of this sort of
communication.
*Many* Thanks in Advance,
Mat
