[Simh] Is it possible to simulate the first Vaxen I ever used?

[Johnny Billquist]

On 2020-03-23 15:44, Paul Koning wrote:
> 
> 
>> On Mar 23, 2020, at 10:34 AM, Dan Gahlinger <dgahling at hotmail.com> wrote:
>>
>> ...
>> I remember they opened the chassis a number of times to show off that bar, the part was indeed labelled "FUBAR", it was the source of some laughs.
> 
> FUBAR is the name of a 780 CSR (in the UBA: failed unibus address register); perhaps it was used in the 730 as well.  I'm sure the engineers got a kick out of being able to sneak that acronym past the writers and managers.

I think the FUBAR was only in the DW780. I think I looked at some point 
around the 11/750 and couldn't even find a FUBAR there.

But anyway, the DW780 FUBAR would not be worthy of a label inside the 
cabinet, so I think this FUBAR Dan saw must have been something else...

>> ...
>> there was what I'd call a bug in the vms on that system.  you could rename a .dir that had files within it to say .dat then delete the .dat if you set it /nodirectory, and all the files within the dir would still use up disk space, even though there was now no longer any way to work with them. so you'd have this missing disk space basically forever, still counting against the users quota.
>> I know because I did that at least twice.
> 
> VMS is like Unix: directories are name to inode number maps (not called "inode"; I forgot the correct name).  A file doesn't need a name.  I remember RSX had an explicit way to reference a file by its number, don't remember what VMS did.

File ID, or FID for short.
And I think it's possible to refer to files by ID in VMS as well. 
Especially from programs, but it might have been possible from DCL as 
well, with various tools.

But anyway, yes, files can get lost. That is one task of fsck (under 
Unix), VFY (under RSX), or ANALYZE (I think it's under VMS). Finding 
lost files and enter them into some known directory.

I wouldn't even call this a bug. Files-11, which is what VMS and RSX 
uses, do not reference count files. You can have multiple entries for 
them in different directories, just like any hard link in Unix, but 
removing such entries have no bearing on the existence of the file or 
not. And likewise, you can also delete a file while still having 
directory entries that refers to it.
It's part of the design of the Files-11 file system.

And that is why file IDs have two numbers. The first one is the 
equivalent of the inode number. The second number is a generation 
number, so that you don't confuse files when a file ID is reused, since 
you can retain, and refer to files by file ID.

   Johnny

-- 
Johnny Billquist                  || "I'm on a bus
                                   ||  on a psychedelic trip
email: bqt at softjar.se             ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol