Re: nsswitch reviewer wanted

In the list that shall not be cc'd to, on Tue, Nov 01, 2005
at 04:15:07PM -0800, Brooks Davis wrote:
> Michael Bushkov disk some interesting work for use under the Google
> Summer of Code and I'd like to see the appropriate parts committed.
> Unfortunately, this isn't an area I have great depths of knowledge so
> I'm hoping someone one else is interested in working on this.

I realise that you weren't asking for comments, but I took a quick look
at http://www.rsu.ru/~bushman/nsswitch_cached/nss_cached.patch and have
some.  I'll send this to the original lists too.

o There aren't nearly enough comments.  Granted, I'm not a C aficionado,
  but it's ~ line 3200 of that patch before we come to a new non-trivial
  comment, and there aren't many after that.

o cached/config.c has magic numbers in create_def_configuration_entry(),
  which probably belong as #defines in config.h instead.  I'm not sure
  what style(9) says about that though, so am happy to be ignored.

o There is a single mention of a ssh_hostkeys cache in
  include/nsswitch.h, and no code to implement it.

o On line 15448, there is a whitespace nit.  Also, in this area, are we
  sure that there is no benefit in continuing to key by euid/egid if
  perform_actual_lookup is enabled; can we send the euid/egid with the
  lookup request?

o A user should be able to invalidate one of their caches (e.g.,
  "cached -i hosts" to flush their hosts cache).  Root should be able
  to do it for all users with a single command (e.g., "cached -I hosts"
  to flush all hosts caches).

o The manual for cached.conf is unclear over whether it's OK to name
  an "unknown" cache in cached.conf.

o The location of cached.conf is defaulted to /usr/local/etc in
  cached/cached.c and should be changed on commit.

Ceri
-- 
Only two things are infinite, the universe and human stupidity, and I'm
not sure about the former.			  -- Einstein (attrib.)