Re: 1036 and back references

From: John Stanley (stanley@peak.org)
Date: Thu Jul 01 2004 - 00:24:46 CDT


Henry Spencer (henry@spsystems.net):

>That's "RECOMMENDED", actually.

It still isn't "REQUIRED".

>And essentially all implementors read it
>as a very strong recommendation, as Charles said -- something to mess with
>only if there is specific and compelling reason.

That's "valid reasons", and it doesn't even have to be compelling, just
valid.

>> SHOULD is a suggestion...

>No, it's a good deal stronger than that.

Not according to RFC2119, which is where the word is defined. Of course,
a "strong" suggestion is still a suggestion, and even a "strong"
suggestion is still not a requirement.

>It's something you are expected
>to do unless unusual circumstances dictate otherwise.

Perhaps on YOUR planet it is unusual to have valid reasons to do things,
but on our planet it is quite common, and on our planet, RFC2119 says
all you need to have is a valid reason.

>There is no reason
>to have a standard key word just to indicate a suggestion.

Of course there is: to differentiate between situations that are
suggestions and those that are requirements.

But you need to take this up with the authors of RFC2119, who seemed
quite content to create a key word that marks a recommendation.




This archive was generated by hypermail 2.1.7.