- From: Henrik Frystyk Nielsen <frystyk@w3.org>
- Date: Wed, 28 Apr 1999 10:33:14 -0400
- To: "Larry Masinter" <masinter@parc.xerox.com>, <ejw@ics.uci.edu>, "Yoram Last" <ylast@mindless.com>
- Cc: "WEBDAV WG" <w3c-dist-auth@w3.org>
At 15:06 4/26/99 PDT, Larry Masinter wrote: >> >My belief is that HTTP interoperability testing for PUT and >> >DELETE has been superficial. > >What I meant was that while we gave people a chart to fill out >for PUT and DELETE with "No (not implemented), Yes (implemented), >Tested (against two independent implementations)", I'm not sure >exactly what they tested, and whether they tested error cases >and the handling of response codes, and whether there was a clear >matching of expectiations of what the state of the resources might >or might not be after the operations were completed. So is the case for all other features of HTTP/1.1 listed on the form. It is not more likely to catch a problem if you try millions of times if you still don't happen to try the specific thing that triggers the problem. I believe we found the serious problems for PUT and DELETE just as we did for other HTTP methods. >> In the later case, there is plenty examples of HTTP/1.1 working just fine, >> see for example >> > http://www.w3.org/WinCom/NoMoreLostUpdates.html > >which is fully implemented and supported in libwww, jigsaw, and amaya. > >This isn't part of the HTTP/1.1 spec, though. And it requires some amount >of support for etags in association with documents that are being PUT. Sure it is - it's plain old HTTP/1.1 - no extensions what so ever and it is all described in the spec. Henrik -- Henrik Frystyk Nielsen, World Wide Web Consortium http://www.w3.org/People/Frystyk
Received on Wednesday, 28 April 1999 10:36:45 UTC