Compound operator chaining does not support interspersed dots

[CosmicToast]

Consider the following object:
{"package": {"foo": "bar"}}

Then consider the queries seeking to get out bar, in jq and jaq respectively.

In jq: .package.foo and .package.["foo"] both return bar.
In jaq, .package.foo returns bar, but .package.["foo"] complains about an invalid token.
.package["foo"] works as expected, but is non-obvious (relative to .["package"]["foo"]; note that .["package"].["foo"] is similarly valid in jq but not jaq).

[wader]

For reference why it was added in jq this comment probably summaries it jqlang/jq#1699 (comment). For strict jq compat it would be good to support it, if not maybe make jaq produce clearer error message about it? currently: Error: expected key <...inline code...> unexpected token

[CosmicToast]

From a UX perspective, .foo.bar and .foo.["bar"] being equivalent makes sense, since it's symmetrical.
You write in the .foo.bar format first, and then simply add [" and "] if you realize you need those for compatibility reasons (in my case, +@ at the start). Needing to take extra steps like removing the . as well is not symmetrical and therefore surprising, so if it's to stay that way I would think it should also warrant a compatibility note and a warning around the part of the manual that talks about [] notation and operator chaining.

[01mf02]

Bonjour @CosmicToast,

thanks for reporting this. I think that #423 should address this.

Greetings to Alsace!