1. There are various things tests that can just have intentions added
into them, like filters and such like, add intentions to these
2. Start thinking about being able to negate steps easily, which will
lead on to a cleanup of the steps
This enables people to enter things using the mouse to paste for
example, plus possible other things.
As an aside it also answers my query regarding `fillIn` for testing,
nothing needs to be actually _typed_ anymore! Doh
This just makes sure that if multiple services are registered with unique service addresses that we don’t blast back multiple CNAMEs for the same service DNS name and keeps us within the DNS specs.
Previously `api-double` usage in ember would require a bunch of `fetch`
requests to pull in the 'api double', this had a number of disadvantages.
1. The doubles needed to be available via HTTP, which meant a short term
solution of rsyncing the double files over to `public` in order to be served
over HTTP. An alternative to that would have been figuring out how to serve
something straight from `node_modules`, which would have been preferable.
2. ember/testem would not serve dot files (so anything starting with a
., like `.config`. To solve this via ember/testem would have involved
digging in to understand how to enable the serving of dot files.
3. ember/testem automatically rewrote urls for non-existant files to
folders, i.e. adding a slash for you, so `/v1/connect/intentions` would
be rewritten to `/v1/connect/intentions/`. This is undesirable, and
solving this via ember/testem would have involved digging deep to
disable that.
Serving the files via HTTP has now changed. The double files are now
embedded into the HTML has 'embedded templates' that can be found by
using the url of the file and a simple `querySelector`. This of course
only happens during testing and means I can fully control the 'serving'
of the doubles now, so I can say goodbye to the need to move files
around, worry about the need to serve dotfiles and the undesirable
trailing slashes rewriting. Winner!
Find the files and embedding them is done using a straightforward
recursive-readdir-sync (the `content-for` functionality is a synchronous
api) as oppose to getting stuck into `broccoli`.
It will allow the following:
* when connectivity is limited (saturated linnks between DCs), only one
single request to refresh an ACL will be sent to ACL master DC instead
of statcking ACL refresh queries
* when extend-cache is used for ACL, do not wait for result, but refresh
the ACL asynchronously, so no delay is not impacting slave DC
* When extend-cache is not used, keep the existing blocking mechanism,
but only send a single refresh request.
This will fix https://github.com/hashicorp/consul/issues/3524
John Cowen