Andre,

Andre Koopal wrote:

On Sun, Oct 23, 2005 at 05:25:04PM +0200, Daniel Roesen wrote:

On Sun, Oct 23, 2005 at 12:21:09PM +0200, Marco d'Itri wrote:

2 - "e-mail" field of the IRT object
Why would you want to hide the "e-mail" field of the IRT object by 
default (forcing the use of -B to get it) ?

Because of morons writing scripts which send mail to every address they
see.

Those morons will just add -B to their scripts.

Which reminds me, why is the hint to use -B in the output of whois. That makes
it very easy to find out.

IIRC, the idea was never to hide the attributes filtered by default. The idea was to avoid people doing:

$ whois -h whois.ripe.net 1.2.3.4 | grep @

And e-mailing a bunch of useless addresses.

We include the comment about -B because this was the first time that the default output of the Whois server was changed to modify the contents of objects. I thought it was very important that people be able to know how to get the original, unmodified objects.

A few data points:

Date Total queries -B queries Total IP's -B IP's 2005-10-23 2139305 58973 2.8% 45069 1143 2.5% 2005-10-24 2237340 72880 3.3% 51970 1346 2.6% 2005-10-25 2569724 170948 6.7% 49852 1521 3.1% 2005-10-26 2562303 98482 3.8% 52478 1526 2.9%

A relatively small percentage of queries actually uses the -B flag, and these queries come from a relatively small percentage of IP addresses. I also looked at the counts of objects returned, and found them to be roughly similar.

(The number "-B" queries is actually an overcount, because I just looked for "B" anywhere in the query string, but a quick look shows that almost all occurrences of "B" are for the flag. The number of IP's is an undercount, because we get a lot of queries from www.ripe.net, and I didn't convert these to the original client IP address. This is for both the total and the -B.)

The message that I take from this is that when you put data in the database, you can assume that most users will get the default output.

--
Shane