On 5/3/24 11:05, Fred Cisin via cctalk wrote:

> "Remembering his conversation at NCC with Marc McDonald about File Allocation 
> Tables in his unfinished, large, and never-released 8-bit MIDAS operating 
> system, Paterson decided that the FAT scheme was a better way to handle disk 
> information than the way CP/M did it." 

Link-list file allocation was hardly new back then.  CDC had been doing
that since the mid-1960s (cf. SCOPE RBR, RBT, FNT FST, etc.). I suspect
other mainframe operating systems using that scheme may even pre-date that.

One thing that I liked about the CDC approach is that you could use
certain pre-defined file names (INPUT OUTPUT, PUNCH) and they would be
disposed of appropriately at end-of job.   Any other "permanent" files
had to be explicitly attached to the job, giving permissions, passwords,
cycles, etc.

Any temporary files were created just by reference and were deleted at
the end-of-job unless explicitly saved as "permanent" files.

None of this IBM "DD" stuff.

--Chuck


Reply via email to