Marco d'Itri <m...@linux.it> writes: > I do not think that inn2 is going to be a problem, because it has been > built with --enable-largefiles for quite a few releases (and yes, > switching was a bit traumatic for the 32 bit systems involved).
> OTOH, inn will be affected. But inn systems are small enough nowadays > that users can just rebuild the database. I'm afraid the problem with INN is worse than the history and overview databases. The on-disk CNFS header contains time_t, so if you're using CNFS as your spool storage, the whole article spool becomes unreadable. It's essentially exactly the same problem that we had with LFS, and I'm really regretting not switching everything on-disk to uint64_t or the like a long time ago. (But there was never a point after we introduced CNFS when that was going to be easy.) -- Russ Allbery (r...@debian.org) <https://www.eyrie.org/~eagle/>
