One World, One Wait-for

Posted by GregHiggins on 11-Jun-2015 17:12

OE 11.5.1  I am looking at OpenEdge.Net.ServerConnection.ClientSocket:WaitForResponse ( i as integer ) and for the life of me, I can't figure out what this is for: wait-for 'U2':u of mhSocket. Have you been running your tests in non-event driven sessions? Code which worked in 11.5 now spews: 

[15/06/11@16:39:42.281-0400] P-001784 T-001556 1 4GL SOCKET CONNECT: TIME=46
[15/06/11@16:39:42.284-0400] P-001784 T-001556 1 4GL SOCKET WRITE: TIME=0; SIZE=33006
[15/06/11@16:39:42.295-0400] P-001784 T-001556 1 4GL -- Encountered an input-blocking statement while executing a user-defined function or non-void method: 'Execute' that is invalid within the current runtime context. (2780)
[15/06/11@16:39:42.295-0400] P-001784 T-001556 1 4GL -- ** ABL Debug-Alert Stack Trace **
[15/06/11@16:39:42.295-0400] P-001784 T-001556 1 4GL -- --> WaitForResponse OpenEdge.Net.ServerConnection.ClientSocket at line 298 (OpenEdge/Net/ServerConnection/ClientSocket.r)
[15/06/11@16:39:42.295-0400] P-001784 T-001556 1 4GL -- Execute OpenEdge.Net.HTTP.Lib.ABLSockets.ABLSocketLibrary at line 186 (OpenEdge/Net/HTTP/Lib/ABLSockets/ABLSocketLibrary.r)
[15/06/11@16:39:42.295-0400] P-001784 T-001556 1 4GL -- Execute OpenEdge.Net.HTTP.HttpClient at line 139 (OpenEdge/Net/HTTP/HttpClient.r)
[15/06/11@16:39:42.295-0400] P-001784 T-001556 1 4GL -- Execute OpenEdge.Net.HTTP.HttpClient at line 94 (OpenEdge/Net/HTTP/HttpClient.r)

...

Are we no longer supposed to execute execute?

OpenEdge Development - Forum

All Replies

Posted by Peter Judge on 12-Jun-2015 07:52

That's so that we don't wait for the socket to time out before returning control. The ReadResponseHandler in the ClientSocket class will APPLY 'U2' when told to do so by the event args (SocketReadEventArg's :ReadComplete property).
 
We've tested in just about all ABL clients (GUI, batch, AppServer) but I've seen this before when you're calling Execute  from within a  UI trigger. To avoid/workaround, use the other Execute();
 
  def var oReq as IHttpRequest.
 def var oResp as IHttpResponse.
  Def var oClient as IHttpClient.
 
  oClient = ClientBuilder:Build():Client.
 oReq = RequestBuilder:Get('http://www.peg.com'):Request.
 
  oResp = ResponseBuilder:Build():Response.
 
  oClient:Execute(oReq, oResp).
Now you pass in the response to be populated (which is actually what the oResp = Execute(oReq) does).
 
-- peter
 
[collapse]
From: GregHiggins [mailto:bounce-GregHiggins@community.progress.com]
Sent: Thursday, 11 June, 2015 18:13
To: TU.OE.Development@community.progress.com
Subject: [Technical Users - OE Development] One World, One Wait-for
 
Thread created by GregHiggins

OE 11.5.1  I am looking at OpenEdge.Net.ServerConnection.ClientSocket:WaitForResponse ( i as integer ) and for the life of me, I can't figure out what this is for: wait-for 'U2':u of mhSocket. Have you been running your tests in non-event driven sessions? Code which worked in 11.5 now spews: 

[15/06/11@16:39:42.281-0400] P-001784 T-001556 1 4GL SOCKET CONNECT: TIME=46
[15/06/11@16:39:42.284-0400] P-001784 T-001556 1 4GL SOCKET WRITE: TIME=0; SIZE=33006
[15/06/11@16:39:42.295-0400] P-001784 T-001556 1 4GL -- Encountered an input-blocking statement while executing a user-defined function or non-void method: 'Execute' that is invalid within the current runtime context. (2780)
[15/06/11@16:39:42.295-0400] P-001784 T-001556 1 4GL -- ** ABL Debug-Alert Stack Trace **
[15/06/11@16:39:42.295-0400] P-001784 T-001556 1 4GL -- --> WaitForResponse OpenEdge.Net.ServerConnection.ClientSocket at line 298 (OpenEdge/Net/ServerConnection/ClientSocket.r)
[15/06/11@16:39:42.295-0400] P-001784 T-001556 1 4GL -- Execute OpenEdge.Net.HTTP.Lib.ABLSockets.ABLSocketLibrary at line 186 (OpenEdge/Net/HTTP/Lib/ABLSockets/ABLSocketLibrary.r)
[15/06/11@16:39:42.295-0400] P-001784 T-001556 1 4GL -- Execute OpenEdge.Net.HTTP.HttpClient at line 139 (OpenEdge/Net/HTTP/HttpClient.r)
[15/06/11@16:39:42.295-0400] P-001784 T-001556 1 4GL -- Execute OpenEdge.Net.HTTP.HttpClient at line 94 (OpenEdge/Net/HTTP/HttpClient.r)

...

Are we no longer supposed to execute execute?

Stop receiving emails on this subject.

Flag this post as spam/abuse.

[/collapse]

Posted by Simon L. Prinsloo on 13-Jun-2015 11:46

Error 2780 occurs most likely because you do not start with the " -IOEverywhere 1" startup parameter.

Add that and the problem should go away.

Posted by Laura Stern on 13-Jun-2015 13:40

Good thought, but not so.  -IOEverywhere 1 is the default in version 11.  Without this, (IOEverywhere 0) you cannot use any IO-blocking statement in a function or non-void method.  It can't be the 1st one, or a stacked one.  Doesn't matter if it is for a modal window (dialog box) or a non-modal one.  

With -IOEverywhere 1, most of these restrictions have gone away, but one remains. This is if you are trying to use a stacked IO-blocking statement for a non-model window, over an existing .NET-style WAIT-FOR and you are in a function or non-void method.  Apparently that is the case here.  

Posted by GregHiggins on 13-Jun-2015 15:16

Wow! Laura hits the nail on the head from almost 1100 miles away.

My standard GUI client expects a .Net form (if none is provided, it installs itself into the system tray -- which is the current case) and runs a corresponding .Net style wait-for. In the past 10 or so years, serving as the core of several dozen applications, this is the first time there has ever been a problem with it.

This thread is closed