On 2/12/2019 4:09 PM, John Levine wrote:
In article <a956ca35-6264-6bd7-38c3-0816a34f7d98@ix.netcom.com> you write:
-=-=-=-=-=-
On 2/12/2019 12:33 PM, John Levine wrote:
In article <1A1FCA40-9172-4FCF-AC8B-2A4A1FE3E11A@verisign.com> you write:
should look into requesting an update of UTS#46 to add the “Armenian dot” or in the protocol itself (e.g. a mapping solution)? No -- the problem is that you need different mappings for different input languages. See the message I just sent. What about URLs that are in a document or database? Their IDNs should be U-labels, not random text that looks sort of like U-labels. The time to clean stuff up is when its entered and you have some idea of the ccontext.
That's wishful thinking. (Or, that horse has left the barn). URLs are any string that happens to resolve on the local browser and then pasted into a document. You can verify that today it even includes uppercase Greek, for example. In a recent discussion someone put it that labels should be valid IDN U labels once they are "in the system". Now, when you make a DNS lookup the label you submit is "in the system". When it's on the side of a bus, to use the other extreme, it's not. Now, where do you draw the line for where "the system" starts? I don't think you can include all HTML documents or HTML mail message or any other place a URL can and does exist today. A./