Re: passwd processing broken in amd64 (time_t probably)

["Julio M. Merino Vidal" <jmmv%NetBSD.org@localhost>]

On Wed, Jan 14, 2009 at 11:19:57PM +0000, Christos Zoulas wrote:
> In article <20090114142752.GA16713%NetBSD.org@localhost>,
> Julio M. Merino Vidal <jmmv%NetBSD.org@localhost> wrote:
> >On Wed, Jan 14, 2009 at 02:03:35PM +0000, Julio M. Merino Vidal wrote:
> >> 
> >> Effectively, hardcoding a '*version = 1' in _pw_opendb before returning
> >> makes getpwent work again.  So it looks like the version field is not
> >> being stored in the database properly.
> >
> >If this is effectively the problem, how does the following look?
> 
> Looks almost the same with what I have in the branch. But it needs more
> work to parse both the old and the new version correctly. I forgot to
> merge my changes. How does it look now?

I'm in the process of trying it, but after a quick look at the code, you
don't seem to check the current version of the database.  What happens if
someone runs pwd_mkdb with -u against an old database?  I guess you will
set the version in the file but only update a single record in it, thus
leaving it inconsistent.

-- 
Julio M. Merino Vidal <jmmv%NetBSD.org@localhost>