Roger Peppe

To update on this, it seems to me like that attribute parser in `internal/attrs.go` is somewhat too lenient. For example, consider this attribute: ``` @foo(a=one b=two, c="three" four) ``` That's...

As a data point, I checked the `go` command's behaviour in this regard: - Go accepts and reads symlinks locally, including when using a directory replace directive and in vendor...

@seh > The restriction on embed may turn out to be onerous, in that one could write a CUE file that embeds a JSON file sitting right next to it...

@seh > > do you know how Go manages to avoid that difficulty, by any chance? > > In [the _rules_go_ Bazel ruleset](https://github.com/bazelbuild/rules_go), both the `go_library` and `go_binary` rules accept...

@seh Do I have it right that you're happy with CUE's current behaviour with respect to symbolic links?

For the record, the executive summary of current (and proposed future) behaviour with respect to symlinks is as follows: **We allow and respect symbolic links _except_ that we reject/ignore them...

Here's an initial list of uses of deprecated APIs as reported by `staticcheck`: ``` cmd/cue/cmd/add.go:177:8: cue.Build is deprecated: use cuecontext.Context.BuildInstances. The use of Instance is being phased out. (SA1019) cmd/cue/cmd/add.go:255:23:...

Here's another example, maybe related, which had a similar effect, but only with the old evaluator: ``` import "strconv" x: *strconv.Atoi(y) | "1" y: x ``` With the old evaluator...

> In which case the only possible result (I think) is `a: 5, b: 5`. ... but clearly, if `a` is 5, then `b` must be `1` (because of the...

> @rogpeppe also brought up the suggestion that a `==a` validator could be an alternative way of spelling: > > ``` > b4?: concrete(a) > ``` FWIW I'm not sure...