Russ Allbery <rra@stanford.edu> wrote on Tue, 2 Apr 2013 at 11:44:22 -0700 in <8738v8ojqx.fsf@windlord.stanford.edu>:
Given that they are worthless programatically, any argument that they are based on a dead API, or that programattic consistency is required, etc., etc. is not compelling. If they are worthless to software, let's make them useful to humans, please.
And this, I think, is the best argument for changing them, and I'm personally mildly inclined to agree, *provided* that it's the last time we ever change them.
Well, I'm just very reluctant to make a firm commitment like that. As I suggested earlier, in the extremely unlikely event that we get an overwhelming outpouring of feedback that the change was wrong (I suggested 10x the comments we had previously received over the past 5 years, compressed into the space of one month), then we would be, I think, churlish to not consider reverting it. Again, I think that is extremely unlikely, but setting a policy that prohibits our reconsidering an error, if it is clear we have made one, would be a mistake. I'd be OK with an advisory commitment though. --jhawk@mit.edu John Hawkinson p.s.: in re Date: headers, ha! Good point. I guess I mean Received: headers, which do seem to have alphabetic abbrevs. But yeah, it's a lot less useful than it used to be. Of course, then there are all these machines that give their time in UTC which I find personally frustrating, not because I have to do arithmetic to figure out the local time (though that's annoying too), but because I have to figure out the offset is...