Ditto :) On 4/10/07, Ⓙⓐⓚⓔ <[EMAIL PROTECTED]> wrote:
thank you John, it always bothered me!On 4/10/07, John Resig <[EMAIL PROTECTED]> wrote: > > Personally, I think this fix should just be removed - especially > considering that we don't even support versions of Firefox, that old, > any more. > > --John > > On 4/10/07, Dan G. Switzer, II < [EMAIL PROTECTED]> wrote: > > > > >I vaguely remember there being some issue with older versions of Firefox > > >that this solved, but don't quote me on that. > > > > Oddly enough, Ajaxian just had reference to a blog entry that talks about > > Prototype's fix for this: > > > > /* Force "Connection: close" for older Mozilla browsers to work > > * around a bug where XMLHttpRequest sends an incorrect > > * Content-length header. See Mozilla Bugzilla #246651. > > */ > > if (this.transport.overrideMimeType && > > (navigator.userAgent.match(/Gecko/(d{4})/) || > > [0,2005])[1] <2005) > > headers['Connection'] = 'close'; > > > > As you can see, they only do the "Connection: close" header if you're > > running a Gecko version in a certain year. I'd imagine they're doing this to > > fix other issues--such as the issue Jim is pointing out. > > > > While I know that generally Browser sniffing is frowned upon, I can see why > > they're doing it in this case--because there doesn't appear to be any kind > > of DOM sniffing trick you can use. We could do: > > > > // Make sure the browser sends the right content length > > if ( xml.overrideMimeType && !document.all ) > > xml.setRequestHeader("Connection", "close"); > > > > That might fix the problem, but I wonder if sending the "Connection: close" > > string causes issues in other browsers outside of IE. > > > > -Dan > > > > > -- Ⓙⓐⓚⓔ - יעקב ʝǡǩȩ ᎫᎪᏦᎬ
