3 ms·

These types of things happen all the time behind the scenes. The entire domain registration system is very loosely coupled as it relates to transfers - too loos

These types of things happen all the time behind the scenes. The entire domain registration system is very loosely coupled as it relates to transfers - too loose imo, and breakage is really easy. We had a situation last week where a registrar made a small change to their registrar name and it threw our whois parsers for a loop. From a casual view, it looked like the other end had cut us off, and a quick deep dive made the parsing issue immediately apparent. Domain transfers was an afterthought to the multi-registrar model and was designed and implemented as policy (i.e. ICANN), not as a technical process (i.e. IETF). I wish we'd done a better job in 1999.

rhizome 15y ago
Your programmers might want to spend some time implementing some (more) error checking and logging.
- freshhawk 15y ago
  There is no standard format for whois data, it's just free form text. All whois parsers are fragile and error prone by definition. When i was working in that field a few years ago best practice was to make some guesses, log everything and have a ton of manual checks.
  - astrodust 15y ago
    There's even fewer standards for how to present the address in the WHOIS data. You leave something open to the end user and they will always get very, ah, creative.
- freejack 15y ago
  By casual view, I mean from the customer side. The change triggered an alarm pretty quickly and from the back-end the cause was pretty obvious and quickly fixed.
  - freejack 15y ago
    hah - as I wrote this, the exact some issue cropped up. Was fixed again within minutes. :)
- astrodust 15y ago
  Generally each registrar has their own format, but as there's thousands of registrars, some of which have various formats for different types of registrations, there's a confusion of possible layouts in what is generally assumed to be only for record-keeping purposes, not data-interchange. If you've ever looked under the hood at a WHOIS parser you'll see it's filled with nice, neat code, a series of extensions, and then a whole pile of hacks to deal with hundreds of edge cases.
- lambda 15y ago
  The Whois protocol, RFC 812, does not specify the result format returned. The protocol is actually incredibly simple; you connect to port 43 via TCP, send a query delimited by a CRLF, the server sends back a plain text, human readable response. Now, many registrars use a format that has lines that look like "Name: Value". But not all do. And those that do will frequently include other text, including terms of service for using their WHOIS service or providing instructions. And some don't use a "Name: Value" format, or they allow for multi-line indented values, or whatnot. Some seem to use "%" to delimit informational lines, as if it's a comment characters, while others don't bother to use anything. So, parsing whois is a bit more like scraping the web than like parsing a real file format. It can break sometimes.
  - rhizome 15y ago
    I understand, and while I do have experience in text processing I would not want to take on a project like this. The number of people responding about this raises the question, though: why is this being reinvented so much? My impression is that everybody writes their own parser, which smells like NIH syndrome.

rhizome 15y ago

Your programmers might want to spend some time implementing some (more) error checking and logging.

freshhawk 15y ago
There is no standard format for whois data, it's just free form text. All whois parsers are fragile and error prone by definition. When i was working in that field a few years ago best practice was to make some guesses, log everything and have a ton of manual checks.
- astrodust 15y ago
  There's even fewer standards for how to present the address in the WHOIS data. You leave something open to the end user and they will always get very, ah, creative.
freejack 15y ago
By casual view, I mean from the customer side. The change triggered an alarm pretty quickly and from the back-end the cause was pretty obvious and quickly fixed.
- freejack 15y ago
  hah - as I wrote this, the exact some issue cropped up. Was fixed again within minutes. :)
astrodust 15y ago
Generally each registrar has their own format, but as there's thousands of registrars, some of which have various formats for different types of registrations, there's a confusion of possible layouts in what is generally assumed to be only for record-keeping purposes, not data-interchange. If you've ever looked under the hood at a WHOIS parser you'll see it's filled with nice, neat code, a series of extensions, and then a whole pile of hacks to deal with hundreds of edge cases.
lambda 15y ago
The Whois protocol, RFC 812, does not specify the result format returned. The protocol is actually incredibly simple; you connect to port 43 via TCP, send a query delimited by a CRLF, the server sends back a plain text, human readable response. Now, many registrars use a format that has lines that look like "Name: Value". But not all do. And those that do will frequently include other text, including terms of service for using their WHOIS service or providing instructions. And some don't use a "Name: Value" format, or they allow for multi-line indented values, or whatnot. Some seem to use "%" to delimit informational lines, as if it's a comment characters, while others don't bother to use anything. So, parsing whois is a bit more like scraping the web than like parsing a real file format. It can break sometimes.
- rhizome 15y ago
  I understand, and while I do have experience in text processing I would not want to take on a project like this. The number of people responding about this raises the question, though: why is this being reinvented so much? My impression is that everybody writes their own parser, which smells like NIH syndrome.