[Info-vax] INDEXF.SYS

[Johnny Billquist]

On 2012-01-28 16:14, Johnny Billquist wrote:
> On 2012-01-28 15:14, hb wrote:
>> On Jan 28, 12:58 pm, VAXman- @SendSpamHere.ORG wrote:
>>
>>> It knows then the file is in a directory but this is NOT maintained
>>> if ANALYZE/DISK/REPAIR is used.
>>
>> True, the back link points to the directory the file IS in and not the
>> one the file WAS in. But that's a result of repairing the disk. The
>> statement that INDEXF doesn't know the directory is incorrect.
>> ...
>
> I think that VAXmain just illustrated that this was not true.
> The back link still pointed to the old directory, even though the file
> now actually was placed in [SYSLOST]. Ie. the backlink is *not* updated,
> and thus is actually incorrect in this case.
>
> And thus, also, it is true that INDEXF doesn't know the directory the
> file is in. It does, in atleast this case, only know where it was
> created. I don't know if renaming the file to be in another directory
> will update the backlink. Could anyone check?

Belay that. I might have been blind.

> As a small side comment, this was one of the important changes between
> ODS-1 and ODS-2. The others I can think of the the change to directories
> to not be 9+3 characters, the move away from Radix50, and that
> directories always should be sorted. But there other smaller changes and
> improvements as well.
>
>>> %ANALDISK-W-LOSTHEADER, file (9443,4,0) MISSING_FILE.TXT;1
>>> not found in a directory
>>
>> It would be nice if analyze would print the back link, here. At least
>> this would make it obvious, whether all the files moved into SYSLOST
>> belonged to a single directory or to multiple ones.
>
> Right. But you can check that later, when the files are in [SYSLOST] as
> well.

Or not, it would appear. Me wrong.

>>> I think that the OP is trying to figure out a [SYSLOST] directory
>>> full of
>>> files. Since the backlink is now that of the [SYSLOST] directory, that's
>>> not going to help. The proper thing to do would first be ANALYZE/DISK to
>>> find the files not found in a directory. Then collect the backlinks from
>>> the headers of these files BEFORE doing an ANALYZE/DISK/REPAIR. That way
>>> there is a way to create some relationship amongst the files
>>> collected by
>>> the /REPAIR.
>>
>> It seems always wise to do an ANALYZE/DISK/NOREPAIR and save or
>> retrieve some data/information before "corrupting" the disk with /
>> REPAIR.
>
> Did I read something wrong. The example that VAXman posted looked like
> the backlink had not changed when the file was entered in [SYSLOST].
>
> /me confused

Oh! I might have been missing a detail. [MISSING_DIRECTORY] had file id 
9419,4 guessing from the backlink of the MISSING_FILE.TXT. Did [SYSLOST] 
not exist at this point?
Because the backlink after analyze is 9419,5. I missed the bump of the 
revision number on the 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