If you think (1) is enough, I don't think it is necessary to harden
the post script. I agree with Carsten on the fact that if a / is
entered, the script can exit with an error explaining that the
password should contains some of this character.
With the small sed hack given previously, don't you think it is enough ?
On Fri, Aug 16, 2013 at 5:10 PM, Carsten Aulbert
<carsten@???> wrote:
> Hi
>
> On 08/16/2013 03:59 PM, Christian Marillat wrote:
>> 1) I don't think user put a / in a password often.
>
> ACK
>
>>
>> 2) All cases I've seen (2) was with generated password.
>>
>> 3/ A solution is to use pwgen without -s and maybe with more characters
>> (12).
>>
>
> If (2) is true, than 3 should help :)
>
>> "pwgen -s" is :
>>
>> ,----
>> | s, --secure
>> | Generate completely random, hard-to-memorize passwords. These should only be used for machine passwords, since otherwise it's almost guaranteed that users will simply write the password on a piece of paper taped to the monitor...
>> `----
>>
>> What do you think ?
>
> pwgen 12 1 shoud be good enough(TM)
>
> I'm not too familiar with failures in postinst scripts (other than these
> are annoying), but possibly just a simple test if the string contained a
> slash would be in order to quit with an error along with some
> documentation what the user needs to do.
>
> But then, "my" patch is pretty minimal and is just for escaping a slash
> which is possibly contained within the password (please note, I have not
> looked carefully, if the variable is reused later on).
>
> I think your solution is ok, but might fail for existing systems
> (unfortunately).
>
> What do other people think?
>
> Cheers
>
> Carsten
>
