- From: Julian Reschke <julian.reschke@gmx.de>
- Date: Tue, 27 May 2008 15:20:48 +0200
- To: Anne van Kesteren <annevk@opera.com>
- CC: Maciej Stachowiak <mjs@apple.com>, Laurens Holst <lholst@students.cs.uu.nl>, public-webapi@w3.org
Anne van Kesteren wrote: >> - library code that acts on an existing XHR instance but did not create >> it, thus has no way what headers already have been set > > Can you point out which libraries have issues with this? It seems pretty > weird that one library creates an object, invokes open() on it, and then > lets another library set the request headers, and then let yet another > set some more headers and invoke send(). It's a common error to make if the code that calls the library is not sure whether the library is going to set a header or not. With the FF3 behaviour, you can set for instance Content-Type, invoke the library method that computes the body, and still be sure that the Content-Type header that gets send will always be valid. Not with IE (the specified behavior). >> (the last one IMHO is the most important one, and the one that really >> requires new functionality) > > I disagree. I know you do. Please don't complain to the HTTP WG about missing guidelines how to handle broken requests when it is this kind of API which actually makes this kind of brokenness likely (again, see <https://jersey.dev.java.net/servlets/ReadMsg?list=users&msgNo=1346>). BR, Julian
Received on Tuesday, 27 May 2008 13:21:39 UTC