Re: Two head revisions?

[Derek R. Price]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Jim Hyslop wrote:
> Derek R. Price wrote:
>>> I also got a copy of the archive prior to the second 1.4
>>> commit.
> Sure
>>> enough, it had a header with "head: 1.3", no 1.4 metadata, but
>>> had a 1.4 change text.  All 3.5 original revisions were created
>>> on the same day, so I can't pull anything older from storage.
>>> :(
>
> Was the change recent enough that the people who committed 1.3 and
> the first 1.4 may remember anything useful, such as whether either
> commit was forced, whether there were any errors or warnings, and
> so on?

It's over a year old, so probably not.  I've talked to the user who
recently encountered the corruption but not the user who created the
original file, but that's a good point and I'll put in a request, just
in case Jane remembers anything.

> Other than what you've mentioned, was there anything else unusual
> about the file? In your original message, you predicted what the
> file probably looked like. How close was your prediction to the
> actual?

My prediction was pretty exact.  I couldn't predict the 1.3 diff
contents since they had been recalculated and overwritten by the
second commit (maybe it would have been possible if I had thought
about it longer, but I didn't try), and it turned out to have an empty
diff, meaning the original 1.4 revision was a force-commit of the same
data from 1.3 with a new log message.

> What is the likelihood that someone manually modified the ,v file?

I'm told that these guys run a very strict shop and that the
possibility that someone manually modified the file is so low that I
shouldn't even consider it and that anyone there attempting something
like that should be in mortal fear of ending up in the street via a
forceful boot to their posterior (I've attempted a loose translation
of the French used in explaining the situation to me... :).

Of course, in conjunction with the force-commit, it looks
superficially like a user might have tried ineptly to rid themselves
of the revision with the wrong log message after using a force-commit
to get a revision with the correct log message into the file, and I'm
not certain the user in question heard the same French phrasing I was
privy to regarding the potential consequences of such an attempt.

> Out of curiosity, does the validate_repo script pick up the
> original corruption?

Yes, it does.  Revisions 1.1 and 1.2 cannot be checked out of the
original corrupt file.  Log also does not work.  Checking out the head
revision and tagging the head revision both work, which is probably
partly what allowed the corruption to go unnoticed for so long.

Regards,

Derek
- --
Derek R. Price
CVS Solutions Architect
Get CVS support at Ximbiot <http://ximbiot.com>!
v: +1 248.835.1260
f: +1 248.835.1263
<mailto:derek@ximbiot.com>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.1 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFEWQHULD1OTBfyMaQRAiTaAKCShTGx0UT98fRA5/dhrctszVxwbQCeNo6R
kOesPrrxir5JyHxlyJ6J8Tw=
=W1WB
-----END PGP SIGNATURE-----