The Python bsddb linkage downgrade to db4.5 also breaks SpamBayes, FWIW. I was able to recover (or apparently so) by using db4.6_dump and db4.5_load; I don't know whether this will work on all Berkeley DB 4.6 files or not, but for anyone else who's getting broken by this, that's what I'd try first on the affected files (keeping a backup of the originals, of course).
Why is this bug still assigned to apt-listchanges? As far as I know, potentially any Python application that uses bsddb to load and save data and expects that the file format will not randomly become unreadable by the next version (a reasonable assumption, I'd think...) is broken by this. That would indicate assigning the bug to Python, no? I see that the Python maintainer, Matthias Klose, has reassigned the bug, claiming that it is not a bug in the Python packages, and that it has to be handled in apt-listchanges. This sounds like nonsense; it would have to be "handled" in potentially every application that used Python Berkeley database access, not to mention that there isn't an obvious way to "handle" it at all! No application could have anticipated the linkage downgrade, AFAIK, and even afterwards, there's no good way for the application to get its data out to do any kind of recovery from the change, since its entire ability to read the file has disappeared---unless I'm grossly misunderstanding the nature of things. The assumption "files written by this (packaged) library will continue to be readable by the next version of the package" seems quite reasonable to me. I'd expect such a heavy backwards-incompatibility as breaking that assumption to at least merit a news entry---which, ironically enough, would be displayed by apt-listchanges before it happily horked itself by making itself unable to read its own previously-written database file by upgrading the Python package anyway. ---> Drake Wilson -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
