Disable sha1 cache busting for HTML Imports

[morethanreal]

Imagine a config that includes a.html and b.html:

a.html:

<link href="b.html" rel="import">
<!-- register custom element x-a -->

b.html:

<!-- register custom element x-b -->

When karma generates context.html, it appends all paths with ?sha1. The above config would throw an error where x-b is registered twice, because b.html was loaded twice, once from context.html and once from a.html since the path was different.

Since using HTML imports to resolve dependency is a very common pattern, perhaps karma should not cache bust resources loaded with HTML Imports?

[pkozlowski-opensource]

@morethanreal I see the problem, but I'm not sure if making Karma aware of HTML imports is the best strategy. The thing is that Karma processor would have to scan all the HTML files, parse out URLs to the imports and then exclude those files. It is doable but sounds a bit hairy.

But I think we can make all those things working soothly if partials (imports) are served by Karma but not included by default. This is possible with the syntax for files described here: http://karma-runner.github.io/0.12/config/files.html. In your case it could look something like this:

files: [
 'a.html'
 {pattern: 'b.html', included: false}
],

I hope that this solves the immediate problem but if there are any enhancements that we could add, I would be happy to look into those. Would the above work for you?

[pkozlowski-opensource]

@morethanreal oh, I've just bumped into your comment here: Polymer/polymer#513 (comment) which basically says the same thing as I'm suggesting above :-)

One again, I'm not sure if scanning all the HTML files to detect HTML import URLs is the best strategy for Karma as it could slow down tests....

[dignifiedquire]

I don't think this is something that karma should be doing.