registry backup (and upgrade --force)
cal at macports.org
Fri Apr 8 12:07:15 PDT 2016
On Wed, Apr 06, 2016 at 03:08:09PM +0200, René J. V. Bertin wrote:
> That doesn't teach me much new
> %> bunzip2 < /opt/local/var/macports/registry/registry-corrupt.db.bz2 > kk.db
> %> /opt/local/bin/sqlite3 kk.db
> %> .load MacPorts-svn/src/cregistry/macports.sqlext
> %> pragma integrity_check;
So your registry wasn't broken, apparently?
> I don't know if it helps, but here are 2 backtraces I got after
> sending a SIGKILL (not SIGQUIT as the log pretends). They're pasted in
> chronological order, but they were preceded by an unknown number of
> other attempts to "get back in", for which I didn't get a crash
I think I have previously had situations where the database seemed to be
rather slow and re-creating fixed it. I did not investigate further,
though, but I suspect you may have hit a similar case.
> I wonder if the "DeleteColumnNames" function above doesn't mean that
> somehow all invalid data has been stripped from the "corrupt"
> database, which would make it valid but probably useless.
I doubt that. SQLite would rather throw an error than loose data.
Loosing data is just about the worst thing a database can do.
More information about the macports-dev