On 16 July 2017 at 10:28, Paul Gilmartin <[email protected]> wrote:
> (From RFC 822: > addr-spec = local-part "@" domain ; global address ) ... > If the local-part contains special characters, it must be surrounded by > quotation marks. If there are no special characters, the quotation > marks are optional. But note that many non-alphanumeric characters are not "special" in this context, and so do not need to be quoted. For example + and & are not special, so Bob&Carol+Ted&[email protected] is valid. > Mappings may make domains equivalent. For example, my domain > may be written as "AIM.com" or "AOL.Com" Mappings may also make local-parts equivalent. Gmail, for example, ignores embedded periods when processing inbound mail. My employer's inbound email handling applies some kind of search to the local-part. As it happens, there is currently no other Tony in the email list at this domain, so Tony is enough to reach me. That will surely change one day, but what will the algorithm be? Reject, ignore, stick to the existing rule and send to me...? > Some servers (not all) accept userID"+"subaccount as a local-part, > delivering mail to userID and allowing rules in userID's MUA to > sort messages into folders. But any RFC-compliant mail *sender* must permit the + character in the local-part. This is a pet peeve. Many web sites have a similer piece of bogus Javascript that rejects all kinds of perfectly valid addresses. I like to use the + convention, but I'd say around 60% of sites I deal with reject it. The worst silently ignore the +, or replace it with a blank, and let the next downstream component fail on it, also perhaps silently. I have had helpdesks tell me I must change my email address in order to deal with their company. Would they do the same if their phone system had a bug such that my area code wasn't accepted?* Most of the big vendors' mail products get this stuff right, but even a few of them have quirks. In my experience Gmail is near perfect, MS Outlook almost as good, IBM/Lotus Notes not so great, and > ... all challenges for DFSORT; some not algorithmically solvable. The syntactic validity of an email address is algorithmically solvable, but the processing and policies of various recipient servers and organizations is of course, not. Hence no amount of algorithmic normalization can reliably tell you if two email address are equivalent or not. Tony H. * Actually I had this problem around ten years ago. Trying to call a number in the UK from a landline here in Canada. "Cannot complete your call as dialled" recording. The operator (does anyone remember the notion of telephone operators?) tried with the same results, and everyone told me the UK number was invalid. I got someone else in the UK to call it, and they got through. Eventually persuaded my local operator to call the UK operator and have them try it. Went through with no trouble. I opened a ticket with repair service (which they didn't want to accept), and eventually they got back to me with "a switch programming error in Toronto". How do you call an operator in 2017...? ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
