Re: pwd_mkdb(8) is still broken to me

[christos%astron.com@localhost (Christos Zoulas)]

In article 
<70f62c5e0901262323r3a2c5320x65bc51bac83e46e4%mail.gmail.com@localhost>,
Masao Uebayashi  <uebayasi%gmail.com@localhost> wrote:
>If I build nbpwd_mkdb on 32-bit time_t host, it generates "VERSION 0"
>*pwd.db files, which can' be read by libc correctly.  This is a fix
>against pwd_mkdb(8) to generate "VERSION 1" *pwd.db files.
>
>        http://www.netbsd.org/~uebayasi/pwd_mkdb-always-version-1-patch.txt
>
>A question is, is there any case where "VERSION 0" is really needed
>and generated?  I think "VERSION 0" handling is broken.  Moreover,
>what is VERSION 0 and VERSION 1?  What is the version of "no VERSION
>string" *pwd.db?  These conditions and specifications are vague.

version 0 or no version means 32 bit time_t
version 1 means 64 bit time_t

new libc can read both; old libc can read the former. If the new libc
cannot read version 0 files, then it is broken.

christos