I need to create a delphi application where when it\'s started the server is started as well and starts sending messages immediately, but I haven\'t found an example or tuto
This example uses a Delphi 2009 VCL application with a main form, which contains only one visual component, a TMemo named “MemoLog”.
Client and server are both started in the FormCreate event. Note that the client code does not handle connection loss, but this can be implemented with a separate re-connect loop within the thread.
procedure TServerPushExampleForm.FormCreate(Sender: TObject);
begin
ExampleServer := TMyPushServer.Create;
ExampleServer.DefaultPort := 8088;
ExampleServer.Active := True;
ExampleClient := TMyPushClientThread.Create('localhost', 8088,
MemoLog.Lines);
end;
The server code uses a TIdTCPCustomServer subclass which waits for a random time and then sends a string to the client.
function TMyPushServer.DoExecute(AContext: TIdContext): Boolean;
begin
Result := inherited;
// simulate hard work
Sleep(Random(3000));
AContext.Connection.IOHandler.WriteLn(
'Completed at ' + TimeToStr(Now), IndyTextEncoding_UTF8);
end;
The client code uses a TThread subclass to run asynchronously without blocking the main VCL thread. It contains a private TIdTCPClient instance, and periodically tries to receive a string from the connection.
...
S := TCPClient.IOHandler.ReadLn(IndyTextEncoding_UTF8);
...
Below is the full code for the example main form.
unit Unit1;
interface
uses
IdCustomTCPServer, IdTCPClient, IdContext,
SysUtils, Classes, Forms, StdCtrls, Controls;
type
TMyPushClientThread = class(TThread)
private
TCPClient: TIdTCPClient;
FLog: TStrings;
public
constructor Create(AHost: string; APort: Word; ALog: TStrings);
destructor Destroy; override;
procedure Execute; override;
end;
TMyPushServer = class (TIdCustomTCPServer)
protected
function DoExecute(AContext: TIdContext): Boolean; override;
end;
TServerPushExampleForm = class(TForm)
MemoLog: TMemo;
procedure FormCreate(Sender: TObject);
procedure FormDestroy(Sender: TObject);
private
ExampleClient: TMyPushClientThread;
ExampleServer: TMyPushServer;
end;
var
ServerPushExampleForm: TServerPushExampleForm;
implementation
uses
IdGlobal;
{$R *.dfm}
procedure TServerPushExampleForm.FormCreate(Sender: TObject);
begin
ExampleServer := TMyPushServer.Create;
ExampleServer.DefaultPort := 8088;
ExampleServer.Active := True;
ExampleClient := TMyPushClientThread.Create('localhost', 8088, MemoLog.Lines);
end;
procedure TServerPushExampleForm.FormDestroy(Sender: TObject);
begin
ExampleServer.Free;
ExampleClient.Terminate;
ExampleClient.WaitFor;
ExampleClient.Free;
end;
{ TMyPushServer }
function TMyPushServer.DoExecute(AContext: TIdContext): Boolean;
begin
Result := inherited;
// simulate hard work
Sleep(Random(3000));
AContext.Connection.IOHandler.WriteLn(
'Completed at ' + TimeToStr(Now), IndyTextEncoding_UTF8);
end;
{ TMyPushClientThread }
constructor TMyPushClientThread.Create(AHost: string; APort: Word; ALog: TStrings);
begin
inherited Create(False);
FLog := ALog;
TCPClient := TIdTCPClient.Create;
TCPClient.Host := AHost;
TCPClient.Port := APort;
TCPClient.ReadTimeout := 500;
end;
destructor TMyPushClientThread.Destroy;
begin
TCPClient.Free;
inherited;
end;
procedure TMyPushClientThread.Execute;
var
S: string;
begin
TCPClient.Connect;
while not Terminated do
begin
S := TCPClient.IOHandler.ReadLn(IndyTextEncoding_UTF8);
if not TCPClient.IOHandler.ReadLnTimedout then
begin
TThread.Queue(nil,
procedure
begin
FLog.Append(S);
end);
end;
end;
TCPClient.Disconnect;
end;
end.
(From https://mikejustin.wordpress.com/2014/04/19/indy-10-tidtcpserver-server-side-message-push-example/)
The way that Indy works is to have the client (TidTCPClient) connect to the server (TidTCPServer) and then exchange data between them, back-and-forth until the connection is terminated either willfully or by premature disconnect.
I am only referring to the actual Indy TCP components here and not to the way you see your applications.
At the application level you might consider an application the server app and another the client app but both can/may contain both TidTCPClient and TidTCPServer components with which they communicate with other apps. This means that the server app can initiate a connection to a client app via the server app's TidTCPClient component and the client app will receive the connection via its TidTCPServer component. This would be a possible solution but keep in mind that generally clients are dynamic and ever changing while servers are usually static and as such it will be a mission to keep track of where clients are. Too many headaches and too much work as well.
So I think it is better to have clients keep track of their rarely changing servers and as such it is better to have a TidTCPServer component for the server app and have it wait for client connections before it starts to send messages.
So to implement; your clients would have to constantly try to connect to the server at regular intervals until it finds the server. The server can then send as many messages as it wants until asked to stop or until premature disconnect in which case the cycle will be restarted. There are ways in Indy to keep track of client connections and you can keep an internal list of the clients through those means. This makes more sense. It is the way that most client-server apps work. Just think of Skype and any Web Server. The clients contacts the server and receives data if needs be.
At the server side:
At the client side:
There are many examples for this type of operation. You must try first and if you struggle you can always ask questions specific to the problem you are having.