channel subscription rules (was: Re: [ticker-dev] Latest draft)

[Ian Lister]

On Mon, 19 Apr 2004, David Arnold wrote:
>i think everyone is now used to dealing with systems that take the
>approach of requiring precise use of capitalisation, and those that
>have decided that such precision is unnecessary.

I don't think they are. In fact I'm damn sure they're not :-)

Your average computer user doesn't deal with a great deal of computer
systems that require precise use of capitalisation. DOS/Windows systems
traditionally don't require it anywhere and the main way it has potential
to leak in is from Unix systems, specifically on the Internet. However,
even most commonly used systems on the Internet (i.e. web and mail) don't
particularly care about it: neither DNS nor mail usernames are case
sensitive, and by the time people need to specify directory paths in URLs
they're more often cutting and pasting or just clicking than having to
type from scratch.

In fact the only place a normal non-Unix user is likely to come across
case sensitivity is in passwords, and I seem to recall at least some
versions of Microsoft products (things like AD, not just the old toy-like,
completely insecure versions of Windows) use case-insensitive passwords.

But even aside from the pseudo-reasoning and speculation on how the other
half lives, my own experience dealing with technical users in a Unix (or
at least Unix-aware) environment (i.e. DSTC) has pretty firmly convinced
me that if usability is even a small priority for Tickertape it should be
case insensitive.

>with regard to decomposition, i suppose Ian has the best knowledge of
>Unicode amongst us?  Ian, what would you suggest is most likely to
>annoy the fewest people?

I'm far from an expert on the subject, but I'll venture a
slightly-better-than-guess opinion. Unless we are going to completely
ignore the issue, I think recommending canonical decomposition is
extremely important if we ever deal in non-ASCII group names (as hopefully
we should). I also very strongly recommend compatibility decomposition,
even more so than case folding. The reason for this is that canonically
equivalent character sequences are completely indistinguishable (without
looking at the code points used to represent them), and that compatibility
equivalent character sequences are only subtly different (and may not
appear at all different in many fonts). On the basis that a user should be
able to look at a group name and reproduce it elsewhere I think it's worth
using compatibility decomposition.

Ian