Top caching Questions

List of Tags

I have a Flash project, and it has many source files. I have a fairly heavily-used class, call it Jenine. I recently (and, perhaps, callously) relocated Jenine from one namespace to another. I thought we were ready - I thought it was time. The new Jenine was better in every way - she had lost some code bloat, she had decoupled herself from a few vestigial class relationships, and she had finally come home to the namespace that she had always secretly known in her heart was the one she truly belonged to. She was among her own kind.

Unfortunately, Flash would have none of that. Perhaps it had formed an attachment. Perhaps it didn't want Jenine to be decoupled. Either way, it clung to the old, perfect version of Jenine in its memory. It refused to move on. It ignored her (function) calls. It tried to forget her new, public interfaces. Instead, every instance of Jenine that it constructed was always a copy of the old version, down to its classpath:

var jenineInstance:Jenine = new Jenine();
trace( getQualifiedClassName(jenineInstance));
// Should print: com.newnamespace.subspace::Jenine
// Prints: com.oldnamespace.subspace::Jenine
// Ah, young love!

We fought. I'm not proud of some of the things I said or did. In the end, in a towering fit of rage, I deleted all references of Jenine completely. She was utterly, completely erased from the system. My cursor fell upon the "Empty Trash" menu option like the cold lid of a casket.

I don't think Flash ever recovered. To this day it still clings to the memory of Jenine. Her old, imperfect definitions still float through my project like abandoned ghosts. Whenever I force Flash to compile, it still lovingly inserts her into my movie, nestling her definition in amongst the other, living classes, like a small shrine. I wonder if they can see her.

Flash and I don't really talk anymore. I write my code, it compiles it. There's a new girl in town named Summer who looks almost identical to Jenine, as if someone had just copied her source-code wholesale into a new class, but Flash hasn't shown any interest. Most days it just mopes around and writes bad poetry in my comments when it thinks I'm not looking.

I hope no one else has had a similar experience, that this is just a singular, painful ripple in the horrifying dark lagoon that is the Flash code-base. If, by some fluke chance you have, or you have any idea how to erase whatever damn cache the compiler is using, please, please help.

Answered By: murzeb ( 448)

Flash still has the ASO file, which is the compiled byte code for your classes. On Windows, you can see the ASO files here:

C:\Documents and Settings\username\Local Settings\Application Data\Adobe\Flash CS4\en\Configuration\Classes\aso

On a Mac, the directory structure is similar in /Users/username/Library/Application Support/


You can remove those files by hand, or in Flash you can select Control->Delete ASO files to remove them.

669
user1684978

Since the upgrade to iOS 6, we are seeing Safari's web view take the liberty of caching $.ajax calls. This is in the context of a PhoneGap application so it is using the Safari WebView. Our $.ajax calls are POST methods and we have cache set to false {cache:false}, but still this is happening. We tried manually adding a timestamp to the headers but it did not help.

We did more research and found that Safari is only returning cached results for web services that have a function signature that is static and does not change from call to call. For instance, imagine a function called something like:

getNewRecordID(intRecordType)

This function receives the same input parameters over and over again, but the data it returns should be different every time.

Must be in Apple's haste to make iOS 6 zip along impressively they got too happy with the cache settings. Has anyone else seen this behavior on iOS 6? If so, what exactly is causing it?


The workaround that we found was to modify the function signature to be something like this:

getNewRecordID(intRecordType, strTimestamp)

and then always pass in a timestamp parameter as well, and just discard that value on the server side. This works around the issue. I hope this helps some other poor soul who spends 15 hours on this issue like I did!

Answered By: Kieran ( 333)

After a bit of investigation, turns out that Safari on iOS6 will cache POSTs that have either no Cache-Control headers or even "Cache-Control: max-age=0".

The only way I've found of preventing this caching from happening at a global level rather than having to hack random querystrings onto the end of service calls is to set "Cache-Control: no-cache".

So:

  • No Cache-Control or Expires headers = iOS6 Safari will cache
  • Cache-Control max-age=0 and an immediate Expires = iOS6 Safari will cache
  • Cache-Control: no-cache = iOS6 Safari will NOT cache

I suspect that Apple is taking advantage of this from the HTTP spec in section 9.5 about POST:

Responses to this method are not cacheable, unless the response includes appropriate Cache-Control or Expires header fields. However, the 303 (See Other) response can be used to direct the user agent to retrieve a cacheable resource.

So in theory you can cache POST responses...who knew. But no other browser maker has ever thought it would be a good idea until now. But that does NOT account for the caching when no Cache-Control or Expires headers are set, only when there are some set. So it must be a bug.

Below is what I use in the right bit of my Apache config to target the whole of my API because as it happens I don't actually want to cache anything, even gets. What I don't know is how to set this just for POSTs.

Header set Cache-Control "no-cache"

Update: Just noticed that I didn't point out that it is only when the POST is the same, so change any of the POST data or URL and you're fine. So you can as mentioned elsewhere just add some random data to the URL or a bit of POST data.

Update: You can limit the "no-cache" just to POSTs if you wish like this in Apache:

SetEnvIf Request_Method "POST" IS_POST
Header set Cache-Control "no-cache" env=IS_POST

I have noticed that some browsers (in particular, Firefox and Opera) are very zealous in using cached copies of .css and .js files, even between browser sessions. This leads to a problem when you update one of these files but the user's browser keeps on using the cached copy.

The question is: what is the most elegant way of forcing the user's browser to reload the file when it has changed?

Ideally the solution would not force the browser to reload the file on every visit to the page. I will post my own solution as an answer, but I am curious if anyone has a better solution and I'll let your votes decide.

Update: After allowing discussion here for a while, I have found John Millikin and da5id's suggestion to be useful. It turns out there is a term for this: auto-versioning. I have posted a new answer below which is a combination of my original solution and John's suggestion.

Another idea which was suggested by SCdF would be to append a bogus query string to the file. (Some Python code to automatically use the timestamp as a bogus query string was submitted by pi.) However, there is some discussion as to whether or not the browser would cache a file with a query string. (Remember, we want the browser to cache the file and use it on future visits. We only want it to fetch the file again when it has changed.) Since it is not clear what happens with a bogus query string, I am not accepting that answer.

Answered By: Leopd ( 33)

Google's mod_pagespeed plugin for apache will do auto-versioning for you. It's really slick.

It parses HTML on its way out of the webserver (works with PHP, rails, python, static HTML -- anything) and rewrites links to CSS, JS, image files so they include an id code. It serves up the files at the modified URLs with a very long cache control on them. When the files change, it automatically changes the URLs so the browser has to re-fetch them. It basically just works, without any changes to your code. It'll even minify your code on the way out too.

165
Morgan Cheng

Is there a standard for what actions F5 and Ctrl+F5 trigger in web browsers?

I once did experiment in IE6 and Firefox 2.x. The "F5" refresh would trigger a HTTP request sent to the server with an "If-Modified-Since" header, while "Ctrl+F5" would not have such a header. In my understanding, F5 will try to utilize cached content as much as possible, while "Ctrl+F5" is intended to abandon all cached content and just retrieve all content from the servers again.

But today, I noticed that in some of the latest browsers (Chrome, IE8) it doesn't work in this way anymore. Both "F5" and "Ctrl+F5" send the "If-Modified-Since" header.

So how is this supposed to work, or (if there is no standard) how do the major browsers differ in how they implement these refresh features?

Answered By: some ( 335)

It is up to the browser but they behave in similar ways.

I have tested FF, IE7, Opera and Chrome.

F5 usually updates the page only if it is modified. The browser usually tries to use all types of cache as much as possible and adds an "If-modified-since" header to the request. Opera differs by sending a "Cache-Control: no-cache".

CTRL-F5 is used to force an update, disregarding any cache. IE7 adds an "Cache-Control: no-cache", as does FF, which also adds "Pragma: no-cache". Chrome does a normal "If-modified-since" and Opera ignores the key.

If I remember correctly it was Netscape which was the first browser to add support for cache-control by adding "Pragma: No-cache" when you pressed CTRL-F5.

Edit: Updated table

The table below is updated with information on what will happen when the browser's refresh-button is clicked (after a request by Joel Coehoorn), and the "max-age=0" Cache-control-header.

Updated table, 27 September 2010

+------------+-----------------------------------------------+
|  UPDATED   |                Firefox 3.x                    |
|27 SEP 2010 |  +--------------------------------------------+
|            |  |             MSIE 8, 7                      |
| Version 3  |  |  +-----------------------------------------+
|            |  |  |          Chrome 6.0                     |
|            |  |  +  +--------------------------------------+
|            |  |  |  |       Chrome 1.0                     |
|            |  |  |  |  +-----------------------------------+
|            |  |  |  |  |    Opera 10, 9                    |
|            |  |  |  |  |  +--------------------------------+
|            |  |  |  |  |  |                                |
+------------+--+--+--|--+-----------------------------------+
|          F5|IM|I |IM|IM|C |                                |
|    SHIFT-F5|- |- |CP|IM|- | Legend:                        |
|     CTRL-F5|CP|C |CP|IM|- | I = "If-Modified-Since"        |
|      ALT-F5|- |- |- |- |*2| P = "Pragma: No-cache"         |
|    ALTGR-F5|- |I |- |- |- | C = "Cache-Control: no-cache"  |
+------------+--+--+--|--+--+ M = "Cache-Control: max-age=0" |
|      CTRL-R|IM|I |IM|IM|C | - = ignored                    |
|CTRL-SHIFT-R|CP|- |CP|- |- |                                |
+------------+--+--+--|--+--+                                |
|       Click|IM|I |IM|IM|C | With 'click' I refer to a      |
| Shift-Click|CP|I |CP|IM|C | mouse click on the browsers    |
|  Ctrl-Click|*1|C |CP|IM|C | refresh-icon.                  |
|   Alt-Click|IM|I |IM|IM|C |                                |
| AltGr-Click|IM|I |- |IM|- |                                |
+------------+--+--+--+--+--+--------------------------------+

Versions tested:

  • Firefox 3.1.6 and 3.0.6 (WINXP)
  • MSIE 8.0.6001 and 7.0.5730.11 (WINXP)
  • Chrome 6.0.472.63 and 1.0.151.48 (WINXP)
  • Opera 10.62 and 9.61 (WINXP)

Notes:

  1. Version 3.0.6 sends I and C, but 3.1.6 opens the page in a new tab, making a normal request with only "I".

  2. Version 10.62 does nothing. 9.61 might do C unless it was a typo in my old table.

Note about Chrome 6.0.472: If you do a forced reload (like CTRL-F5) it behaves like the url is internally marked to always do a forced reload. The flag is cleared if you go to the address bar and press enter.