Re: Exp state on top of dead

[Mark D. Baushke]

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

CARTER-HITCHIN, David, GBM <address@hidden> writes:

> Got it.  It's installed and configured to run on a regular basis.   Only
> one slight thing I'd like to change with the script would be to have 
> a little reason keycode by each file at the end of the report as to why
> the script thinks it's corrupt.  We've got approx 700,000 files in our
> repository and if we ran -v on the whole lot we'd have too much output.

Given that it is possible to have more than one 'corruption' in a given
file, but that not all checks would be performed once some kind of
corruption is detected, I am not certain how useful a keycode would
really be.

Corruptions:
    No revisions found.

    Corrupted in revision: $revsion
        If one revision is corrupted, it is typically impossible to
        reconstruct older revisions on the trunk or newer revisions on a
        branch, so more revisions would also likely be corrupt.

    $name and Attic/$name BOTH exist

    No head revision found.

    In the Attic with non-dead head revision

    NOT in the Attic with a dead head revision

Exec checks:
    The --exec scripts may also indicate via a zero/non-zero return
    value that a particular revision is 'corrupt'. This is typically
    used to do consistency checks for things like conflict markers or
    'binary' files that are not actually valid. They are run against
    every revision of every file in the repository.

A possible way to help isolate your corruptions would be to take the
list of files that are listed as being corrupt and copy them into a new
dummy repository and then rerun the validate_repo script with --verbose
on those set of files. This would let you work on fixing those files in
isolation.

        -- Mark
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.4 (FreeBSD)

iD8DBQFFAaZ4Cg7APGsDnFERAhhmAKCeR+d8UXcAFzEA6pG8G1SmomdQuQCgyKtT
BC1cHJis43/M923eTTuQABE=
=HSQF
-----END PGP SIGNATURE-----