On 2017-09-18 22:39, Ted Lemon wrote:
On Sep 18, 2017, at 4:35 PM, Julian Reschke <julian.reschke@xxxxxx
<mailto:julian.reschke@xxxxxx>> wrote:
Concrete example: a system text editor that defaults to the system
encoding (for historic reasons). Are you saying it's buggy when it
continues to do so?
No, the system that doesn't have UTF-8 as its default encoding is buggy.
Is it the RFC Editor's task to change this?
Now, at this present time. If you are still running Windows XP and
you see some artifacts when interacting with the modern Internet, that's
not our problem.
Once again, dealing with "the internet" is not a problem.
What's a problem is handling these files once they are stored locally.
The paragraph I quoted simply says that UTF-8 BOM isn't useful or
recommended. So it's agreeing with Carsten, and I guess with me. If
we disagree with the consortium on this, that's okay, but we ought to
have a really good reason for disagreeing.
What it says that it's not needed for distinguishing between the various
Unicode encoding schemes (of which, IIUC, we want to get rid of all but
UTF-8 anyway).
IMHO it doesn't address distinguishing between Unicode encoding schemes
and *other* encoding schemes. For which I believe we have evidence that
it *is* useful in practice in some cases. (And yes, I hear you Martin:
UTF-8 can be reliably sniffed; but the applications we tested with didn't).
I'm not saying the the decision to use the BOM was the correct one. It
was reached after a long public discussion though.
Best regards, Julian