Attribute doubts/proposals

[jugaadi]

1. Any reason for not allowing attributes at the top of fields/struct/package identifiers? For multi-line string, putting an attribute at the end of the field affects the readability.
2. Can I use cue inside an attribute as part of its content? This will enable reuse of the content.
3. Should we reserve attribute names based on file types? So @py will always represent python
   (edited)

[myitcv]

@jugaadi - thank you for re-raising this on GitHub discussions.

Any reason for not allowing attributes at the top of fields/struct/package identifiers? For multi-line string, putting an attribute at the end of the field affects the readability.

You can use embedded scalars to make this a bit more readable:

f: {
	@hello(This is
		a story about a very
		big attribute message
		)
	4
}

If there are other situations that feel awkward, please can you share them?

Can I use cue inside an attribute as part of its content? This will enable reuse of the content.

Attributes do not influence the evaluation of CUE. They exist to allow associating meta information with values.

Should we reserve attribute names based on file types? So @py will always represent python

This is a really good question that I think we should discuss in the form of an issue if you don't mind creating an issue. Certain attributes are already "reserved" by virtue of their use in the CUE distribution (e.g. @protobuf). Should the CUE distribution reserve others: possibly? Will we be able to enforce this? Probably not. But still worth discussing.

[jugaadi]

1. It should be applied to the field/struct/package that immediately follows it or the attribute below it.

Output

<a x="Hello World" y=5>Foobar</a>

Input1:

_m: 4
_n: 1
_o: "World"

a: Foobar @xml(x="Hello \(_o)" y=\(m+n))

or

Input2:

_a: {
    x: "Hello World"
    y: 5
}
a: Foobar @xml(_a)

Here I would like to use cue expressions inside the attribute. Or did I misunderstand the usage?

[myitcv]

1. It should be applied to the field/struct/package that immediately follows it or the attribute below it.

That's problematic for the reason I explained above.

Here I would like to use cue expressions inside the attribute. Or did I misunderstand the usage?

From the spec:

Attributes allow associating meta information with values. Their primary purpose is to define mappings between CUE and other representations. Attributes do not influence the evaluation of CUE.

Hence, you cannot use a CUE value in an attribute. Attributes exist outside of CUE evaluation.

[jugaadi]

Thanks @myitcv for your response. Should I create an issue(Feature Request) for the above two?

[myitcv]

By all means create feature requests, but as I've outlined I would be surprised if we ever move to allow attributes to be defined before an identifier; similarly, it's feature of attributes that they exist outside of CUE evaluation.

[jugaadi]

For (3), I have created an issue here - #1019

[cueckoo]

This discussion has been migrated to cue-lang/cue#1009.

For more details about CUE's migration to a new home, please see cue-lang/cue#1078.