Creating a file with timestamp 4780

Peter T. Breuer

Well, the one I am using now is pretty old - but I will take a peek. Nope - only one processor. Yes it would be best to do things that do not depend upon the events - I was just trying to find a way to get unique file names -

Well, I am still glad that John just mentioned *one*. It is nice to get a *little* help or a hint.

I thought the same thing would be accomplished by John's suggestion - if memory serves me correctly, using "fopen" on a file that is already open produces an error and not a type of "process queuing situation" for resources.

OK - will try again - what I meant is something like the Round-Robin approach: where each process gets a little timeslice of the CPU (only one process can hold the attention of the CPU at a time). It was why I was thinking that time() in process A would not be the same as time() in process B. But - as you have mentioned already - that would depend upon the granularity as to what you would actually get. Besides, time() seems to go only as far as "seconds".

See above.

I am a bit biased. From my experience on systems, whenever an APPLICATION blows up, there is information placed into an application log file - the stack, programming code where it stopped, date, time, all sorts of information. This would also be used for debugging purposes. I guess kinda like the dump of the core being used with gdb.

There would also be one application log for system errors buttociated with each process (showing detailed SYSTEM errors) and another general log that would contain all SYSTEM errors.

So, from what I have seen in the past, the file locking mechanism you speak of is buttociated with for the recording of system information - not application.

Just trying to stick with what has worked for others.

Thanks.