Chris Holland: But you're right, this is all a lot of end-user
intervention: it would
be a slightly, err, very painful process of installing a browser
plugin, which is currently very-much of a user opt-in process, and not
something very practical.
[...]. I'm just trying to find ways to leverage a lot of what's
already there.
Yes, I don't like the idea of requiring users to act. But I see what
you were tying to do with reuse.
Interesting thing is the same scheme could be leveraged for local CSS
extensions:
I thought about this as well, but CSS is far less likely to be
duplicated across multiple sites. There are plenty of CSS Frameworks
but I feel that none have picked up enough dominance for this kind of
optimization to be useful.
You do mention what looks like urn schemes and extending this idea to
CSS. I was specifically thinking of javascript because of its
widespread use of libraries/frameworks. Using URN Schemes could let
this repository idea extend to more then just javascript, however I
don't think any other type of resource (CSS, Images, Etc.) have this
unique pattern of "the exact same content being served on thousands of
different domains."
<link rel="local:extension" type="text/css" href="ext:ibdom.
0.2.js" />
To handle users who don't have the "ibdom" javascript extension
installed, developers could add something like this to their document:
(assuming a decent library which declares a top-level
object/namespace):
<script type="text/javascript">
if (!window.IBDOM) {
var newScript = document.createElement("script");
scr.setAttribute("type","text/javascript");
scr.setAttribute("src","/path/to/ibdom.0.2.js");
document.appendChild(newScript);
}
</script>
Although the idea is the same (have a fall back plan if a repository
lookup is ignored or fails), I think this is needlessly complex
compared to just adding a new attribute on the <script> tag. By
extending the script tag you already have fall back behavior to just
download the script from the "src" attribute.
If you take the <link> approach then you're practically requiring that
you need to write fault tolerant code like you showed above, and that
is no fun for web developers.
------
The more I think about it, the more I think this might not necessarily
be a web standards idea. More of a browser optimization, however it
would never take off unless it was standardized. I don't know what to
do about this... CDN Caching, like Google's Hosted Libraries, is more
generic but less optimized. Maybe this is just a special case?
- Joe
