package: shadow

Hi,

currently, when doing d-i installs with DEBCONF_PRIORITY, root and user 
passwords are only asked once (if not preseeded). IMO this is a grave bug, as 
this provides no way to detect typos, so users will choose simple passwords. 
(Or make typos...)  And it's also different from all password prompting user 
interfaces I have seen.

<h01ger> bubulle: are you still of the opinion that it's sane to only ask for 
the rootpw once if DEBCONF_PRIORITY=critical ?
<bubulle> h01ger: yes, but, well, my opinion is maybe not what is to be 
implemented, after all.... I gave my arguments when this discussion occured a 
while ago, I have no new argument pro or against this.
<h01ger> bubulle: i'm strictly against asking for passwords only once. How to 
detect typos that way ? There is no way so people will choose passwords like 
"mate" or "123" :-( If you ask for passwords, you have to confirm them. For 
critical installation mode, $disabled as a password would be much more 
handy :)
<h01ger> bubulle: but we can discuss this nicely at debconf or maybe 
linuxtag/karlsruhe allready ?
<bubulle> h01ger: Sure. I think that, indeed, this decision is among those 
which pertain to the whole d-i team.
<bubulle> As shadow maintainer now (sigh), I will implement what is judged as 
most appropriate by the d-i team, as this feature is only used during 
installs
<bubulle> [...] I *will* deal with that post-sarge...but, again, after taking 
opinions from either the d-i team, or the technical comitee, or by starting a 
flamew^W discussion in -devel
<h01ger> bubulle: you might even argue that it's a debian decision. as 
"ergonomic user interfaces" are demanded by some laws (you are not allowed to 
use unergonomic software) and entering a password only once is against all 
users expectations. - even admins have a right for ergonomic software :-) but 
i absolutly agree with post-sarge and team-decision.
<bubulle> h01ger: yep, the decision about prompting the root pw twice is a 
general design decision, so a "debian" decision (thus, technical comitee, 
again?)


regards,
 Holger

Attachment: pgpLnETbU2XeM.pgp
Description: PGP signature

Reply via email to