- From: Jeffrey Mogul <mogul@pa.dec.com>
- Date: Wed, 24 Apr 96 11:00:15 MDT
- To: hallam@w3.org
- Cc: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
I'm somewhat disturbed by calling CRLF a "NULL Request". This seems to me to open up a lot of opportunities to fall into holes when writing statements like "Every request receives a response code." Plus I don't think it quite captures the problem for the server writer. Server writers are going to get extraneous CRLFs appended to certain content types and they have to be aware of that. I suggest we call it what it is, a KLUGDE (or a hickup). Then we are not likely to commit an error later on by inadvertently refering to "Requests" and including NULL requests by mistake. I think that there should also be a statement to warn server writers not to depend on the additional CRLF. I think you are on the right track here. In RFC791, there is a succinct description of what has become known as the "robustness principle" In general, an implementation must be conservative in its sending behavior, and liberal in its receiving behavior. This was restated in RFC1122, and is generally a good guide for deciding how to approach this kind of borderline-acceptable behavior. I suggest that the HTTP spec say something vaguely along these lines: (1) Don't send extra CR-LFs in HTTP messages (2) The recipient of an HTTP message should ignore CR-LFs (and probably other "whitespace") if by ignoring that whitespace it can convert a syntactically incorrect message into a syntactically correct one without any ambiguity. In other words, you all c a n read this sentence even if it looks like e. e. cummings wrote it. Then we don't need to worry about the complexities of trying to define a null method, just to avoid honoring the robustness principle. -Jeff
Received on Wednesday, 24 April 1996 11:11:42 UTC