Peter Schuller wrote in a message to All:

 PS> I'm currently developing a BBS system, and of course -
 PS> multinode. This means  that there will be multiple threads
 PS> or processes (depending on how I'll  implement it) be
 PS> accessing the same files! Let's take the message-files  and
 PS> file-files for instance. If 4 nodes wishes to read/write to
 PS> the same  message/file-base (thus, the same file), how do I
 PS> coordinate this? Semphores'  not good, considering there
 PS> could be 100 nodes running (unlikely, but 50  maybe).

Use DosSetFileLocks().  If you are working in C++, you can wrap file
locking in an object oriented wrapped.

 PS> Best way I can figure out:

 PS> I create a 'access-handleing-thread', that as above receives
 PS> messages via  queues or pipes (probably queues, since pipes'
 PS> point-to-point), and then send  messages to other threads
 PS> that requested the access. This thread will then  have the
 PS> access until it sends
 PS> "i'm-finished-let-the-next-thread-get-the- access" to the
 PS> access-handleing-thread, that then if there is one, passes
 PS> the  access on. But this too can get complicated.

I'm sure Peter will have something to say on this, since he is responsible
for the MCP/MCP32 components of Maximus.  MCP32 is a DLL which exports a
bunch of services, allowing different instances of Maximus to communicate:

   Entry Index  Name
         1      McpOpenPipe
         2      McpClosePipe
         3      McpSendMsg
         4      McpGetMsg
         5      McpDispatchMsg
         6      McpPeekMsg

Having a master database server daemon might be a reasonable approach if
you are really expecting to support 50 or 100 nodes.  Pipes would be more
appropriate for what you want to do, and they can be networked at some
point in the future if you need that.
 
-- Mike


---