Hi Neil,

Although I never used PHP, by looking at your snippet I recognize that 
this solution is not different from the previous -- both circumvent the 
caching mechanisms between the server and its clients :-(

This solution is worse than the previous one since it requires the 
installation of some server-side software (the PHP script requires a 
special scripting engine).

But I guess that I will have to live with this:
"The Mozilla XMLHttpRequest does not cache at all - you have to handle 
that. The IE implementation caches by default unless told not to by the 
server or when a request header is sent that involves caching (like 
If-Modified-Since)"
...until somebody at Microsoft will dare to fix this bug (in Mozilla it 
is already fixed).

Thanks for your time,
Adrian.



Neil Smith [MVP Digital Media] wrote:
> Yes, set the headers correctly on your server for the directory.
> 
> In PHP you'd use this (then send the XML file contents) :
> 
> header( "Expires: Sat, 1 Jan 2005 01:00:00 GMT" ); 
> header( "Last-Modified: " . gmdate( "D, d M Y H:i:s" ) . "GMT" ); 
> header( "Cache-Control: no-cache, must-revalidate" ); 
> header( "Pragma: no-cache" );
> 
> There's more information about controlling this in IE (client-side)
> here :
> 
> http://jpspan.sourceforge.net/wiki/doku.php?id=javascript:xmlhttprequest:behaviour:httpheaders#conditional_get
> 
> HTH
> Cheers - Neil
> 
> On Fri, 01 Jul 2005 15:01:58 +0300, Adrian Herscu
> <bmf1972@a...> wrote:
> 
> 
>>Any suggestion about how to work around this XMLHTTPRequest caching problem?
>>
>>Adrian.
>>
>>BTW Mozilla does the opposite - it always brings the XML files without 
>>caching (although this has been fixed in developer builds).
>>
>>Adrian Herscu wrote:
>>
>>
>>>Hi Neil,
>>>
>>>Thanks for your fast reply.
>>>
>>>I know, there should be a better terminology.
>>>
>>>By "static" I meant to a file that is not generated using a 
>>>CGI/ASP/JSP/PHP/ISAPI/...
>>>
>>>But all "static" files are updated once in a time. Few examples: CSS 
>>>files are updated in order to bring a new look to the (dynamically 
>>>generated) Web pages, JS files are updated in order to bring new 
>>>behavior to the Web pages, HTML & PDF files are updated because the 
>>>documentation changed, etc., etc., etc.
>>>
>>>Then, why XML files are supposed to remain unchanged for the rest of 
>>>their life?
>>>
>>>By appending such a "random GET string to the URL" it will defeat the 
>>>caching mechanism, and files will always be retrieved. I need to load 
>>>those files only if they are newer than the cached ones. There is a 
>>>special HTTP header that does exactly that: "if-modified-since".
>>>
>>>MS-IE uses this header to bring all ("static") files which are 
>>>_hyperlinked_ from the main HTML file (i.e., scripts, stylesheets, XML, 
>>>XSLT, images, etc.) - because some mysterious reason, MS-XML decides not 
>>>to revalidate it cache.
>>>
>>>Any workaround?
>>>
>>>Thanks
>>>Adrian.
>>>
>>>Neil Smith [MVP Digital Media] wrote:
>>>
>>>
>>>>(Q : If it's static, why does it matter if it retrieves from the
>>>>cache, surely a static file *hasn't* changed, that's what the cache is
>>>>for !)
>>>>
>>>>A : As for all cached files you want to regenerate (e.g. a logging
>>>>image) append a random GET string to the URL, so for example
>>>>
>>>><script language="javascript1.2">
>>>>var randomItem=new Date();
>>>>documentHref=documentHref+'?random='+randomItem.getUTCMilliseconds();
>>>>//    Etc etc
>>>></script>
>>>>
>>>>On Thu, 23 Jun 2005 00:08:23 +0300, Adrian Herscu
>>>><bmf1972@a...> wrote:
>>>>
>>>>
>>>>
>>>>>Hi all,
>>>>>
>>>>>I am using MSXML 3.0 to synchronously load XML documents in the browser.
>>>>>
>>>>>Here is a snippet:
>>>>>
>>>>>xmlHttpRequest = new ActiveXObject( "MSXML2.XMLHTTP.3.0" );
>>>>>xmlHttpRequest.open( "GET", documentHref, false );
>>>>>xmlHttpRequest.send( null );
>>>>>
>>>>>documentHref always points to a _static_ XML file.
>>>>>
>>>>>This generates no HTTP request - the browser retrieves the file from 
>>>>>its cache.
>>>>>
>>>>>Is it possible to make it send a conditional GET?
>>>>>
>>>>>Thanks for your time,
>>>>>Adrian.
>>>>
>>>>
>>>>
>