?

Log in

No account? Create an account

fanf

CRSIDs and email addresses

« previous entry | next entry »
2nd Jun 2009 | 13:09

I've recently written a three page memo about the advantages of our username scheme compared to "friendly name" email addresses in large domains. I have put a copy of the PDF on the web which you can read if you like.

| Leave a comment | Share

Comments {9}

Simon Tatham

from: simont
date: 2nd Jun 2009 13:34 (UTC)

Section 2.9 was particularly well worded, I thought :-)

Reply | Thread

Steve

from: timeplease
date: 2nd Jun 2009 15:13 (UTC)

Yes.

Reply | Parent | Thread

Ted

from: erikofviking
date: 2nd Jun 2009 21:41 (UTC)

All good points. The embarrassing situation of sensitive financial information landing in an undergrad mailbox rather than with the director of finance happened up here in the North earlier in the year.

We took the potentially controversial decision to delete it directly out of the users mailbox in the end. Not something that I was overly happy at having to do.

Reply | Thread

from: Dave Holland [org.uk]
date: 3rd Jun 2009 09:45 (UTC)

Interesting read, thanks. You could mention the lessons learned from the period around 1993 when long CRSIDs of the form djh1008 were issued, and the problems with people misreading 1 as l, 0 as o, etc. Presumably the 1/l problem will become more widespread when enough people have CRSIDs that abc10, abc11 etc become common - or has that point already been reached?

Reply | Thread

Tony Finch

from: fanf
date: 3rd Jun 2009 11:23 (UTC)

I made a deliberate decision not to discuss past CRSID policies in any detail, to keep the document short and focussed. I also omit things like group IDs, and advantages of CRSIDs that aren't relevant to @cam email addresses (e.g. CRSIDs are unique enough that they are often usable on public non-University services).

The current policy, in force since September 1994, is:

  • Personal CRSIDs are between 5 and 7 characters starting with the initials of the person (as known at the time of registration) and using at least two alphabetic characters and terminating with at least one digit.
  • The first digit must be >1 (this is to avoid confusion between 1 and I and 0 and O)
  • For those with more than 6 inititals, we chop the initials of the forenames after the fifth.
  • People with only one name have the initial of that name doubled.
  • In the case of hyphanated forenames, both are used to build the CRSid.
  • Single initial CRSids are used to denote courses.
  • Group resources are alphabetic only.

Reply | Parent | Thread

Arnhem

from: arnhem
date: 3rd Jun 2009 14:28 (UTC)

I'm not sure I'm happy with the bits that say "but it's ok for institutions to have friendly names".

It rather assumes that all the institutions are very small compared to the university as a whole, but some of them aren't.

Reply | Thread

Tony Finch

from: fanf
date: 3rd Jun 2009 15:20 (UTC)

Agreed.

The aim of the document is to discourage foolish people from advocating the use of friendly name addresses in the @cam domain. I guess a lot of the pressure for this could be relieved by supporting friendly name addresses @admin, so we're deliberately emphasizing that the current policy is not against friendly name addresses and in fact actively supports them, for example in the way the managed mail domain interface works.

I avoided a discussion of large departmental domains to keep the document short. It's a bit of a can of worms. There are questions of what categories of people you give email addresses to and what effect that has on the rate you chew through the namespace. In particular, if (like Chemistry) you exclude students, Engineering is less than a third the size of the UAS.

I expect admin will find out that they are large enough that they have significant problems with name clashes. In the past the CS would have tried to lead them away from that kind of trouble, but nowadays it's probably better to avoid seeming too obstructive.

One area in which the instigators of this make-work do not seem interested is to improve the way we use domain names. For example, if we allow large departments to have divisional subdomains (like Physics) then friendly name addresses become more feasible - though of course it has the downside of exposing more organizational complexity.

So yes, that's the kind of nuance and hedging I wanted to avoid in favour of simplified and direct language, so that non-technical senior managers can get a better idea of the considerations behind the status quo, and hopefully they'll think about more than just the surface appearance of email addresses.

Reply | Parent | Thread

cozminsky

What does CRSID stand for.

from: cozminsky
date: 10th Jun 2009 00:39 (UTC)

They sound neat for a domain owner with many users, but I'm still not clear on what it stands for.

Reply | Thread

Tony Finch

Re: What does CRSID stand for.

from: fanf
date: 10th Jun 2009 10:44 (UTC)

"Common Registration Scheme identifier". This is perhaps more obvious to Cambridge people most of whom won't immediately recognize the jargon in the first line of the Introduction but should recognize that it is what CRS stands for.

Reply | Parent | Thread