Re: Verbose data file path reporting

[Fyodor <fyodor () insecure org>]

On Wed, Jun 06, 2007 at 04:04:08PM -0600, David Fifield wrote:
> I have checked into /nmap-exp/soc07/nmap a revision that allows Nmap to
> show which data files it loaded and where it loaded them from. Right
> now, this information is displayed only in verbose mode. Here are some
> examples:

Hi David.  This looks great!  It works in my initial testing.  One
change I would make relates to this situation:

>       # nmap -v -A 192.168.0.0/24
>       Read from /usr/local/share/nmap: nmap-mac-prefixes nmap-os-db nmap-os-fingerprints nmap-rpc nmap-service-probes 
> nmap-services.

For this case where all the files are ready from one directory (which
will be true 99% of the time), I'd rather just print something shorter
like:

"Read data files from: /usr/local/share/nmap"

This way users who don't change data file locations need not deal with
as much output complexity or length.

> The patch doesn't say if a different file name, rather than just a
> different directory, is used. The user has to know, for example, that
> my-services was loaded instead of nmap-services.

Thats OK.  A different filename is only used if they explicity request
it, so the user better know what is going on :).

> This code involves splitting a path into its directory and file
> components. There are POSIX functions dirname and basename that do this,
> but they are exceedingly nasty and hard to use, not to mention not
> available on Windows. I wrote short reimplementations of those functions
> and put them in output.cc. I noticed, though, that a similar problem has
> been solved at least twice before: in main.cc:main to get the nmap
> executable name, and in nse_init.cc:get_filename. Is this something that
> should be moved into utils.cc or nbase?

Sure, please do!  Nbase sounds like a good place for it.  While you
are it it, would you remove the interactive_names nonsense?  I'm
embarassed that feature even exists!

> Is the placement of this
> new information in the output right? It has to be near the end (after
> the port tables) because not all data files have been opened until then.

The beginning would be best, but that doesn't work for the
implementation reasons you noted.  So I think putting the info at the
end as you have done is fine.

Cheers,
Fyodor

_______________________________________________
Sent through the nmap-dev mailing list
http://cgi.insecure.org/mailman/listinfo/nmap-dev
Archived at http://SecLists.Org