Re: bin/46226: underscore character ignored at the end of password

[Matthew Mondor <mm_lists%pulsar-zone.net@localhost>]

The following reply was made to PR bin/46226; it has been noted by GNATS.

From: Matthew Mondor <mm_lists%pulsar-zone.net@localhost>
To: gnats-bugs%netbsd.org@localhost
Cc: 
Subject: Re: bin/46226: underscore character ignored at the end of password
Date: Mon, 19 Mar 2012 08:21:46 -0400

 On Mon, 19 Mar 2012 13:57:44 +0200
 Wladimir Sidorenko <wlsidorenko%gmail.com@localhost> wrote:
 
 > The whole new password was considered including the 9-th character. I beg
 > your pardon for this trouble. Unfortunately I couldn't find any mention of
 > this limitation neither in man page for passwd nor for passwd.conf. A
 > friend of mine told me that this limit could have been mentioned during the
 > installation procedure. But I can't remember now whether it was.
 
 If you installed using the installer, I'm surprised that old was the
 default, however.
 
 But this topic was recently discussed on IRC and it seems that most
 people would prefer sysinst to stop asking which cipher to use in the
 future and default to sha1.
 
 I think that I agree, as crypt(3) is backwards-compatible (supports
 various hash types and can recognize them), and someone who wants to
 generate a password database for an old or specific system can do so by
 manually configuring passwd.conf(5).
 
 However, it seems that for some modern systems the default number of
 rounds is small.  Raising this would however affect login performance
 considerably on slow systems, though.  I wonder if it'd make sense to
 consider using a bogomips type heuristic in sysinst to set a decent
 value...
 -- 
 Matt