cvs commit: src/include pwd.h src/lib/libc/gen getpwent.c src/usr.sbin/pwd_mkdb pwd_mkdb.c

[Jacques A. Vidrine]

On Sun, Apr 20, 2003 at 02:19:55PM -0700, Doug Barton wrote:
> I've already told Jacques that I'd look into the problem with named, but
> more generally, I doubt this will be our only problem child. Has this new
> code been regression tested against any other third party code?

I suppose no more or less than is expected for -CURRENT.

It happens that I don't use BIND8, so I missed this issue.  It
didn't occur to me that applications would use our internal
definitions. (Note the underscores in _PW_KEYBYNAME et. al.)
Moreover, it didn't occur to me that applications would not use
getpwent(3) and friends to access system databases.  (I believe this
is a bug in our BIND installation.)

For this particular issue, as described in the commit, I've reverted
these macros to the previous values in case there is any other code
that (ab)uses them.

As designed, the on-disk format is compatible, so e.g. static binaries
build against FreeBSD 3.x still function with /etc/pwd.db.

[ I may have misunderstood your question.  It's pretty much a no-no to
  commit code with no testing, even to -CURRENT.  Maybe you have
  something more specific in mind. ]

Cheers,
-- 
Jacques A. Vidrine <nectar at celabo.org>          http://www.celabo.org/
NTT/Verio SME          .     FreeBSD UNIX     .       Heimdal Kerberos
jvidrine at verio.net     .  nectar at FreeBSD.org  .          nectar at kth.se