On 5/10/19 12:32 PM, Ted Lemon wrote:
On May 10, 2019, at 2:12 PM, Eliot Lear <lear@xxxxxxxxx <mailto:lear@xxxxxxxxx>> wrote:This is a matter of market power and who really sets the standard. If the choice is between not accepting a buggy implementation and selling product, guess which wins.*cough* produce liability *cough*Honestly, I get that this is how people think, but it’s extremely short-sighted.
So the answer should be strict-only code, but reality.I wish there was a way to publish bugs/issues with implementations (without publicly naming the company) , not as a criticism, but as a way to help the next guy implement. And to help with the next revision of the spec.
With iCalendar for example, there is more than one way to represent the same event. It is ambiguous and you have to examine other implementations objects to see how they do things. Then add code to also accept their way. Every time I try to point it out, it quickly escalates into a 'which way is better / you don't understand'. Which is implementer shorthand for "you change your code, I will not".
A recent Alexa tool sends out appointments with the entire VCALENDAR:DESCRIPTION in the SUMMARY line. Valid, but it really screws up UI's. Technically correct, but time for a new implementation specific switch in the UI code.
And many confuse user experience with data model. Sometimes members attempt to make the protocol represent the methods they need, like an extended API when in some cases the data should be expressed and specified separately.
I think SMTP needs a major overhaul. I am afraid of the noise generated from such a draft.
-- Doug Royer - (http://DougRoyer.US) Douglas.Royer@xxxxxxxxx 714-989-6135
<<attachment: smime.p7s>>
