- From: Getify <getify@gmail.com>
- Date: Tue, 12 Oct 2010 14:17:26 -0500
- To: "public html" <public-html@w3.org>
?> So it seems that this is possible today quite easily: first prime the > browser's cache with all the scripts using one of the techniques described > below, and then when you want to start running the script, insert them one > at a time into the document, using the "onload" event handler to figure > out when to load the next one. > > The "right" way to do cache priming is to explicitly use <link > rel=prefetch>. As I understand it, not all browsers support rel=prefetch; > but it turns out there's a bunch of hacks one can do already to prime the > cache: using an <img> or an <iframe>, for example. So instead of > rel=prefetch, you can use one of those mechanisms. (In some browsers, as > you've seen, <script type="bogus"> is another way to do it. However, that > interferes with the point of the type="" attribute, which is to give the > browser a hint about whether there's any point downloading the file or > not, so I wouldn't recommend using that. It's also non-conforming to > specify the wrong type, since normally it indicates an authoring mistake.) As I just explained in the other response, yes, this "preloading" (or as you call it, cache priming) is *a* solution to the use-case using existing technology. But it's far from ideal. Right now, it's the fallback/hacky solution that LABjs uses for IE/Webkit/Chrome ONLY because no other solution exists. It has several caveats/drawbacks, and it makes LABjs much more complicated than I would like it to be. See the other response for a more thorough explanation of why I assert that it's in the opposite direction of ideal or desired for this use-case. --Kyle
Received on Tuesday, 12 October 2010 19:18:05 UTC