[Top] [All Lists]

Re: Making group syntax do-not-generate

2005-01-14 06:42:12

Keith Moore writes:
1. Mail senders SHOULD NOT generate group syntax.

maybe, maybe not.  groups are useful.

They're not used...

they are used, just not terribly often. but the utility of a feature is not directly related to the frequency of its use.

I agree with the latter sentence.

(Btw, I can think of at least three other reasons why they don't parse groups, so I don't think that presumption is warranted.)

which are?

Some good reasons, some bad:

1. A judgment that it's better for the user to optimize the UI for the simple case, ie. no groups visible.

2. Some people choose to support only what a set of major MUAs generates. (Many people develop with limited time and money, and test all that they support. Sometimes that entails making choices noone likes.)

3. "It's ugly". You know, the RFC2047/2231 argument.

4. Some people test only with a corpus of real-world mail. Me, I have a corpus of 1.6 million messages where groups don't occur even once. In such a setting, it's easy for group-related bugs to slip through.

surely it's harder to read and adhere to the combination of both RFCs X and Y? particularly when they give conflicting advice?

Specious comparison. People aren't really adhering to RFC X in the first place.

but are we going to help the situation by giving them more to read? or to put it another way, where is the barrier?

The primary barrier is IMO in the large number of deployed mail parsers/readers that haven't been tested with groups, and cannot be assumed to work.

or maybe the barrier is in trying to get so many different implementations of a complex protocol (which doesn't support enough features as it is) to act consistently and compatibly with one another?