IEInternals

A look at Internet Explorer from the inside out. @EricLaw left Microsoft in 2012, but was named an IE MVP in '13 & an IE userAgent (http://useragents.ie) in '14

Internet Explorer's Cache-Control Extensions

Internet Explorer's Cache-Control Extensions

  • Comments 5

Some time ago, I wrote a summary of how Internet Explorer’s cache works. At the time, I left out mention of the two cache-control directives introduced by IE5: pre-check and post-check. These directives enable a “background update” mechanism where a cached resource is reused while simultaneously a background revalidation of the resource is performed, ensuring that any updates to the resource will be cached locally before the resource is reused a third time.

MSDN defines these directives as follows:

  • post-check - Defines an interval in seconds after which an entity should be checked for freshness. The check will happen after the user is shown the cached resource, but ensures that on the next visit the cached copy will be up-to-date.
  • pre-check - Defines an interval in seconds after which an entity must be checked for freshness prior to showing the user the cached resource.

The (sadly grainy & typo-ridden) timeline from MSDN shows how this mechanism might be used:

Caching timeline

As it turns out, there are few use-cases on today’s web where a site is willing to show the user potentially outdated content but also believes that it’s important enough to perform a background freshness check.

What’s worse is that, in my experience, no sites sending the post-check and pre-check directives are using them properly. Watching in Fiddler, all use of these directives I’ve seen is as follows:

Cache-Control: no-store, no-cache, must-revalidate, pre-check=0, post-check=0

Where exactly this bad pattern originally came from, I’m not sure, but it’s now replicated in many popular frameworks, and tons of sites send this header to try to forbid caching. If you want to prevent caching, do not include the post-check and pre-check directives. Doing so is completely unnecessary and results in wasted bandwidth and HTTP header processing cycles. In one of the IE7 betas, this pattern even resulted in a somewhat amusing and extremely frustrating bug.

A few other points about these directives:

  • post-check and pre-check must appear as a pair. If only one is present, it is ignored.
  • If both post-check and pre-check are specified and set to 0, both are entirely ignored.
  • post-check must be <= pre-check. If it is not, it will be set = to pre-check.
  • If post-check is set to 0 (and pre-check is >0), a background task will spin up to freshen the cache of the just-downloaded file.
    • So, if you want to double the load on your site for no reason, send Cache-Control: post-check=0; pre-check=1
  • Generally, the pre-check directive is very similar to max-age.
    • However, IE's implementation of max-age takes the Age response header into account. In contrast, the implementation of post-check and pre-check do not. Hence, pre-check is equivalent to max-age only when there is no Age header on the response.
    • IE will stop examining Cache-Control headers for expiration time information after it encounters a header containing a max-age directive. So, if you send multiple Cache-Control headers, post-check and pre-check should appear in the same header as the directive containing the max-age directive.

If you’re using post-check and pre-check on your site, I’d love to hear from you.  Thanks!

-Eric

  • Eric, that's a really interesting feature. Is this an IE only feature? I can't seem to find any official spec's or mentions of pre & post-check outside of MSDN. Few discussions on Mozilla related sites, but nothing conclusive.

  • @igrigorik: These were extension directives added by IE5. As far as I know, no non-WinINET-based browser supports them.

  • we are currently using this extension on http://www.romatoday.it (and the rest of the network)

    it works like a charm, but unfortunatly only on IE.

  • It would be awesome if somebody wrote a spec about this and submitted it to the IETF, so that the extension values end up in the Cache-Control extension registry (see greenbytes.de/.../draft-ietf-httpbis-p6-cache-19.html).

  • @Julian: I think most are hoping that these directives just die.

Page 1 of 1 (5 items)
Leave a Comment
  • Please add 2 and 8 and type the answer here:
  • Post