Re: deprecating Postel's principle- considered harmful

[Adam Roach <[e-mail]>]

On 5/7/19 4:13 PM, Joel M. Halpern wrote:
> I note that both Adam and Barry in their emails talk about this being 
> particularly applicable at the upper layers.


Please don't confuse "at least" with "particularly" in my response. I'm 
just acknowledging that I don't have the expertise to speak to the other 
layers.

I know that lower layers *have* had issues with accepting and attempting 
to interpret, e.g., Christmas Tree Packets with varying degrees of issue 
arising, so the situation is certainly not *isolated* to applications.

/a



>
> On 5/7/19 5:00 PM, Adam Roach wrote:
>> On 5/7/19 3:48 PM, Andrew G. Malis wrote:
>>> I don't agree that poor application programming is a result of the 
>>> Postel principle, it's a result of incompetence or laziness.
>>
>>
>> For better or worse, significant portions of the Internet -- at least 
>> at the application layer -- run on what you're calling incompetence 
>> and/or laziness [1] . The question is: to what degree has Postel's 
>> Principle contributed to this state of affairs; and, if we think it's 
>> a major factor, can we change things so that future protocols don't 
>> suffer from this as much?
>>
>> To be clear, I'm not reading this as trying to put the genie back in 
>> the bottle for already-deployed protocols like SMTP. I read this as 
>> suggesting that maybe future protocols should be a bit more picky 
>> about not accepting messages that are malformed or sequences of 
>> messages that are unorthodox, even if some degree of processing is 
>> technically possible.
>>
>> /a
>>
>> ____
>> [1] More generously, they're probably more the result of things like 
>> cutting corners to meet deadlines and budgets, when the people 
>> cutting corners suffer no consequences for the resulting protocol 
>> pollution.
>>
>>
>