winsock example

[Ben_W_]

Hi, all:

I'm trying to use winsock module to do some inter-process communication. I found that there is very little information from Intel and on the web regarding this. I found a forum topic which is very interesting (http://software.intel.com/en-us/forums/topic/293456). I gained some knowledge by reading that topic. Unfornuatually the attachment in that topic (winsock_example.f90), which contains a winsock example is missing from Intel wbsite. Steve, do you know how and where I can get that exampel file.

Thanks,

Jim

[Steven_L_Intel1]

I'll see if it can be recovered. But why not send Paul Curtis a private message asking that it be reposted?

[onkelhotte]

intel@breault.com wrote:

Hi, all:

I'm trying to use winsock module to do some inter-process communication. I found that there is very little information from Intel and on the web regarding this. I found a forum topic which is very interesting (http://software.intel.com/en-us/forums/topic/293456). I gained some knowledge by reading that topic. Unfornuatually the attachment in that topic (winsock_example.f90), which contains a winsock example is missing from Intel wbsite. Steve, do you know how and where I can get that exampel file.

Thanks,

Jim

Hi Jim,

take a look on this thread:
http://software.intel.com/en-us/forums/topic/291484

There is a link to a GitHub project which I implemented in one of my programs and it works just fine.

Markus

[Ben_W_]

Hi, Steve and Markus:

Thanks for your information. I will contact Paul and Gib for the winsock example file. Markus, I already found your winsock program on GitHub, which gets me started on thinking using windows socket for my project. I was planning using named pipes. But I concerned the compatability between windows named pipes and Linux named pipes as we port my program onto Linux platform in the future. Using TCP/IP will minimize such concern, though we still need come up a TCP/IP fortran solution on Linux platform later.

Thanks,

Jim

[onkelhotte]

intel@breault.com wrote:

Hi, Steve and Markus:

Thanks for your information. I will contact Paul and Gib for the winsock example file. Markus, I already found your winsock program on GitHub, which gets me started on thinking using windows socket for my project. I was planning using named pipes. But I concerned the compatability between windows named pipes and Linux named pipes as we port my program onto Linux platform in the future. Using TCP/IP will minimize such concern, though we still need come up a TCP/IP fortran solution on Linux platform later.

Thanks,

Jim

It´s not my program on GitHub... 

I use it to communicate with some Siemens S7 machines. It works quite good.

Markus

[Paul_Curtis]

The attached winsock_example.f90 illustrates (at least one manner of) how to do socket-based communications from F90.  These routines are used extensively in a SCADA program which communicates perfectly with external networked devices such as PLCs; I do not know whether this would be useful for "inter-process" communications as the OP requested.  In my programs communications code is in separate threads from the main GUI and computational threads so the communications never loads the system; this is why all these routines have been declared recursive.

[Paul_Curtis]

My Winsock-F90 routines have recently been rewritten to use a different set of WinAPI functions, with the result that the communications is greatly improved (much faster, no errors or dropouts), particularly in multithreaded situations.  The update pertains more to WinAPI usage than F90, but anyone who has developed code based on these examples will benefit from an examination of the improved version.  The update file is attached.
 

[DavidWhite]

Paul,

Not sure whether the file uploaded and downloaded correctly.  There is a strange line at 239, which appears to have wrapped two lines into one, with the second part becoming part of the comment at the end of the first line:

                                                 n_sent, & ! number of bytes sent                                                     TRUE,      & ! wait for completion

Thanks,

David

[Paul_Curtis]

Yes, there was a minor typo error in the file.  A corrected version is appended.