[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Design? libshisa doesn't use shishi
From: |
Elrond |
Subject: |
Re: Design? libshisa doesn't use shishi |
Date: |
Sat, 25 Mar 2006 12:50:28 +0100 |
User-agent: |
Mutt/1.5.9i |
On Sat, Mar 25, 2006 at 12:12:51PM +0100, Simon Josefsson wrote:
[...]
> My goal was that Shisa would be a very small and simple user database
> interface library, so it could be used by GNU SASL and other
> non-KerberosV5 packages too. Other packages also need to lookup a
> password for a user, and there is no such generic library on your
> regular Debian box. So that's why Shisa doesn't call any Shishi
> functions.
Ahhh. Getting the point.
> Perhaps Shisa should be split off Shishi completely, although it is
> somewhat krb5-specific right now.
Right. The etype is from shishi.h for example.
> Is this too restrictive? Perhaps we can add a new library that use
> functions from both Shisa and Shishi, which can be used by shishid and
> other Shishi administration tools.
libshisa-util or so. Yeah, that might be an idea.
I couldn't stop myself from looking again at
apply_options() in shisa.c. It still looks like it can be
split out into some tool function taking the full
address@hidden name, etype, password, salt (possibly being
NULL --> default), and string2key params and giving out a
complete Shisa_key.
Not to mention, that apply_options currently has a memleak
on the shishi_key *key. But simply using shishi_key_done()
would be a bad idea either, as dbkey has pointers into that
memory.
It's currently not that important, as shisa is quite
shortlived and frees all memory at exit anyway.
Elrond