Allow overriding base path for UI/API routes; rm --query.prefix

[yurishkuro]

Resolves #745 Provide a means to define a path-prefix for the UI
Resolves #746 Remove --query.prefix option

• depends on Use <base> and config webpack at runtime to allow path prefix jaeger-ui#198 merged

[yurishkuro]

@tiffon this models the compiled index.html from UI assets.

Q: is it possible that babel/webpack would reorder the href and data-inject-target attributes? Because if it does, the search&replace in the query service won't work

[tiffon]

@yurishkuro It should not change the order of the attributes.

[coveralls]

Coverage remained the same at 100.0% when pulling 7a019df on fix-745-support-base-path into 37e6430 on master.

[jpkrohling]

LGTM, but would be awesome to have the user notified when the HTML on disk is not what is being served to clients. Also, it would perhaps be worth adding some piece of documentation about this (when to use/why/how)

[jpkrohling]

Is it possible to add a log statement notifying that what people see in the HTML is not what will be served to clients?

[yurishkuro]

could you elaborate why you think this would be useful? We're modifying the internal file of the app, not something that users provide.

[jpkrohling]

Mainly because it's deviating from what is "expected": if I had a problem with the static resources (like the original problem this PR is solving) and were to check the HTML, I would wonder why it's not the same as what my browser is seeing. A log entry at debug level would suffice, IMO.

[yurishkuro]

well, still not convinced why someone needs to know what happens internally, e.g. we also override the config, and in both cases we only do that when requested via cli arguments.

Let's do it separately, because the constructor doesn't take a logger and the unit tests explicitly verify that nothing is logged, so it's not as trivial a change as it sounds for a feature that seems rather marginal.

[yurishkuro]

Docs would be updated in a separate PR in the docs repo.