[Info-vax] fixing a saveset's attributes: attachment, not ftp

[David Froble]

Stephen Hoffman wrote:
> On 2015-06-22 18:16:40 +0000, hb said:
> 
>> On 06/22/2015 07:05 PM, Stephen Hoffman wrote:
>>> But the implementation of BACKUP here is... broken.   Or we wouldn't
>>> still be having this discussion.  Again.
>>
>> You mean the "solution" /REPAIR is broken, or?
> 
> I mean that BACKUP is being unnecessarily intolerant about its input 
> file handling here.   BACKUP should better follow the classic tenet "Be 
> strict in what you send and tolerant in what you receive."    But then 
> I'm in a particularly charitable mood today, too.   BACKUP should be 
> able to open and to read and verify the integrity a saveset its been 
> presented, and resolving any metadata confusion transparently.  Even an 
> informational here is... well, why even bother the user, if the 
> checksums verified and the BACKUP worked?
> 
> I'd be slightly more forgiving around a tool that wasn't designed with 
> robust saveset file integrity checking, and expressly intended for data 
> recovery from potentially flaky input devices and potentially corrupt 
> files.  To a tool that wasn't so often fodder for these "how do I fix 
> it?" discussions with inexperienced users.   That's before considering 
> that some of the users will be working under pressure to get a server 
> recovered and operational, too.   But then we've been fixing this BACKUP 
> metadata setting for how many decades, yet BACKUP can't manage to open 
> and read and verify the file itself, without needing the associated 
> metadata manually reset via BACKUP /REPAIR or via some DCL or other 
> hackery?
> 
> Or is this particular user interface error so familiar and so large that 
> few notice it anymore?
> 
> 
> 

I would agree that it's rather trivial to get the blocksize from the 
saveset.  The record format might be a bit more of a problem.  After 
all, it is VMS, which has file attributes as a standard feature, and 
it's not expecting too much for BACKUP to need a few things "right".

Now, if BACKUP would look at the saveset, and re-open it with the proper 
blocksize, that would be Ok.  But do you want BACKUP to actually change 
the file attributes?  What if it wasn't a saveset?  Users cvan be rather 
inept.  Could create more problems if the file itself was modified.

Yes, I know the following works, so it maybe could be possible for 
BACKUP to use the file without any attribute changes.

         open F$ for input as file 1%, &
                 organization undefined, &
                 recordtype any, &
                 access read

I'd still want it to look at a few things to insure the file actually is 
a saveset.