You don't want to diff the files (assuming you are not allowing changes on the server end).
Just open the file for appending, which means asking the file size, and setting the position marker to that value before writing.
But you STILL need to keep the data locally, and keep track of what has been put into the master file, and what has not.
FOR EXAMPLE:
Suppose you pick a chunk of 10 minutes, or 10 k bytes.
you write your data to local file "My Local File 1.txt"
When the time or size is exceeded, you switch to writing into "My Local File 2.txt"
You tell your copymaster routine that block 1 is available. (Index into queue).
Your copymaster routine (running separately) gets notified that block 1 is available.
It tries to OPEN/CREATE the network master file.
it Finds the current file size, say 10123 bytes.
It sets the file position to 10123.
It reads file block 1 local file, and writes that data to the master file.
Close the master file.
IF SUCCESSFUL
Remove index 1 from the queue.
Delete local block 1 file (optional)
ELSE
Wait 30 sec and try again.
END IF
IF QUEUE IS EMPTY
Sleep for 10 minutes, or watch for file size > 10k
else repeat with next block index in queue.
