Software Quality Assurance Forums: Grappling with the scope of FileCSVLoadGlobalFile

I am using Silk 2010 and am having some fun grappling with the "global" scope of FileCSVLoadGlobalFile. I have found that multiple VUs actually seem to see the same rows in the file unless I wrap the file access in a mutex. I have this now, and it works well:

Code:

 
  transaction TInit
  begin
    
    // Load CSV and set the delimiter to ","
    hMutex := CreateMutex("SerializeIDList");

    FileCSVLoadGlobal(hUIDFile, UIDFilePath, ",");
    FileGetRow(hUIDFile, POST_Start_User_Row);

Then:

Code:

 
    // Advance to the next row of the file and
    // get the UID and User Name from the current row
    // The application accepts nicknames only up to
    // 20 chars.  When re read them here, we may truncate to 20.
    // The names in the ID-list should be 8+" "+8 at most -- 19
    // characters.
    //
    WaitForSingleObject(hMutex, INFINITE);
    nRow := FileGetNextRow(hUIDFile);
    ThisUserID := FileGetCol(hUIDFile, 2, 20);
    ThisUserName := FileGetCol(hUIDFile, 3, 20);  // may truncate
    
    // restrict to this range
    //
    if (nRow >= POST_End_User_Row) then
          FileGetRow(hUIDFile, POST_Start_User_Row);
    end;
        
    ReleaseMutex(hMutex);

What I want to do is to have different scripts load the same file, but operate over different ranges of lines in the file. Script A will use rows 1001-5000, B 5001-10000, something like that. A and B multiple VUs should run A and B.

So to get this right, I think I need to understand what is the scope that FileCSVLoadGlobal() is global to? Is it the whole controller, regardless of what script? Only global to instances of a given BDH? All instances of the specific file path?

Unlike with the CreateMutex() I don't get to make up my own scope, and the scope rules don't seem to be clearly spelled out.

If anyone can clarify or suggest how to handle this, I would appreciate. Thanks.

--------------------
Gardner

Edited by gardner (12/16/11 03:03 PM)