A stupid Ajax cache problem solution
One of the common problem using Ajax (Flash too) interactions is the cache.
There are a lot of valid PHP, Python, JSP and .NET solutions but browser compatibility is often a question mark.
You could implement easyly and directly a client solution and this is just another proposal.
exactly 123 bytes to solve cache problems and this is the func description:
You could test directly in a loop, a benchmark that many other sites didn't test ...
A script that use Ajax requestes inside a loop is not a good script (I suppose) but this kind of demostration can show You that generated no-cache collisions probability are quite impossible.
A usage example should be this one:
Finally these are some F.A.Q.
Why there is a key and a value and not just a random value to perfom without caching problems ?
- because some server-side code should loop over GET or POST keys and in this way it should know that noCache is a not useful parameter to parse or to check. In other cases a server-side code should consider generated radom value as a key.
Why there is a getTime plus a random value and not just the first one ?
- because some client application should call more than a single request at the same time
Why there is a full stop between getDate and random value ?
- because if client date is modified there are less possibilities that generated value was just used (paranoia style)
I think these F.A.Q. are enought and I hope You'll find this simple function useful :-)
P.S. with ActionScript just change the regExp replacing with
There are a lot of valid PHP, Python, JSP and .NET solutions but browser compatibility is often a question mark.
You could implement easyly and directly a client solution and this is just another proposal.
function noCache(uri){return uri.concat(/\?/.test(uri)?"&":"?","noCache=",(new Date).getTime(),".",Math.random()*1234567)};
exactly 123 bytes to solve cache problems and this is the func description:
function noCache(
uri
// uri string to open
){
return uri.concat(
// concat String prototype,
// the fastest way to produce
// a complete string using multiple values
/\?/.test(uri) ?
// if uri has a query string
"&"
// add last value using & separator char
:
// else
"?",
// add a query string to this url
"noCache=",
// this should be a "cool name" for generated key
(new Date).getTime(),
// the noCache value will be milliseconds
// from 1970/01/01
".",
// plus a dot ...
Math.random()*1234567
// ... and a random value using
// a "big" integer as generator
);
// then this is a return example using uri: http://host.com/mypage.html
// http://host.com/mypage.html?noCache=1166301156233.332083.6663326991
// while this is an example using uri: http://host.com/mypage.html?v0=1&v2=a
// http://host.com/mypage.html?v0=1&v2=a&noCache=1166301168420.631416.7190624559
};
You could test directly in a loop, a benchmark that many other sites didn't test ...
for(var
i = 0, // many loops
max = 10000, // max i value
uri = document.location.href, // this href
obj = {}; // a generic object
i < max; // while i is less than max
i++ // increment the i value
) {
if(!obj[noCache(uri)]) // if obj has not a nocache(uri) key
obj[noCache(uri)] = true; // set them as true
else { // else if obj has just the returned nocache(uri) key
i = max;
alert("noCache doesn't work"); // this method is not so cool
}
};
A script that use Ajax requestes inside a loop is not a good script (I suppose) but this kind of demostration can show You that generated no-cache collisions probability are quite impossible.
A usage example should be this one:
XHR.open("get", noCache(myUri), true);
// or ...
XHR.open("post", noCache(myUri2), true);
Finally these are some F.A.Q.
Why there is a key and a value and not just a random value to perfom without caching problems ?
- because some server-side code should loop over GET or POST keys and in this way it should know that noCache is a not useful parameter to parse or to check. In other cases a server-side code should consider generated radom value as a key.
Why there is a getTime plus a random value and not just the first one ?
- because some client application should call more than a single request at the same time
Why there is a full stop between getDate and random value ?
- because if client date is modified there are less possibilities that generated value was just used (paranoia style)
I think these F.A.Q. are enought and I hope You'll find this simple function useful :-)
P.S. with ActionScript just change the regExp replacing with
uri.indexOf("?") >= 0
Comments
Post a Comment