HTTP/1.1 Feature List Report Summary (original) (raw)

| Forum

Public Reports

It is clear from looking at the data that some people have marked a feature "not implemented" when in fact, the feature did not apply to their implementation, and "n/a" may have been more appropriate.

The Millicent proxy has been implemented multiple times, by different people, in different languages (original prototype in Modula-3, product implementations in C++ and Java) and different physical locations. It is a somewhat specialized proxy, implementing the Millicent Micropayment system (and uses standard HTTP facilities as part of its implementation).

Confidential Reports

Some vendors prefer that their detailed reports remain confidential; the details are rolled up into the full report, but the detailed implementation information is not available publically. Confidentiality is often due to reporting on software not yet available, or for other competitive reasons. Our thanks for the data.

Remaining Features to Test

According to IETF rules, we need 2 interoperable implementations of each feature. We'll feel most comfortable if we can interpret this as two each of clients, servers and proxies, at least where the requirements on proxies differ from clients and servers.

We've not yet tried to figure out exactly for which features the requirements differ (many requirements on Proxies are also true for servers or clients), so the table here significantly overstates the actual amount of testing remaining required to progress the document to draft standard.

Proxies, of course, are the most challenging. There are other proxy implementations for which we do not yet have data, so we can hope testing is better than the data appears. Note that the Millicent proxies are specialized in nature, rather than general purpose caching proxies. Digest, as expected, is in by far the worst state, though implementation proceeds at a good pace. New features to fix old bugs in 2068 are also still a bit of a problem.

We've not yet done extensive analysis on this data; e.g. some features just don't really apply to a given circumstance, and there is certainly errors in the data, from what I've seen.

Anyone who can help in testing those items remaining would be greatly appreciated. From the following, the public testing reports and your own information, you should be able to figure out where your further testing efforts would be most helpful to the community. (e.g. the working group is short a tested implementation, you have implemented the feature, and by testing it against someone elses implementation, you can at least report your implementation tested, and possibly the other implementation as well). By looking at the public reports, you will often be able to find someone who may have implementations you can test against.

Features particularly needing implementation and testing:

• Server implementation of Transfer coding fixes (TE; Henrik Frystyk has a portable client, and Yves Lafon has a server you can easily use for testing your implementations). Note that this feature can significantly improve perceived performance.
• The Trailers header.
• The new version of Digest, client and server and proxy (Ronald Tschalaer has a client, and Dave Kristol has a server that implement the new Digest specification.

See the following list for exact details of features needing testing.

Remaining Features needing testing as of July 21, 1998.

Henrik Frystyk Nielsen, Jim Gettys

@(#) Id:Overview.html,v1.461998/07/2116:44:03jgExpId: Overview.html,v 1.46 1998/07/21 16:44:03 jg Exp Id:Overview.html,v1.461998/07/2116:44:03jgExp