How should hierarchical names (`.` separated) be expanded about uritemplate-test HOT 6 CLOSED

I interpret it the same way. I suspect the python implementation that I maintain expects this to be a literal . in the name of the variable though. Primarily because this is such a subtle line and as you pointed out, there are no tests to catch this in an implementation

from uritemplate-test.

Great. Assuming this is the correct interpretation, what should the result of {;foo.bar} be? I think ;bar=baz makes sense, but since the spec doesn't mention how to expand hierarchical names, I think it would have to be treated like any other name, ;foo.bar=baz which is not very useful.

from uritemplate-test.

Great

To be clear, I have nothing to do with the spec, so just because I interpret it the same way doesn't make us right. :)

what should the result of {;foo.bar} be?

I'd argue that is completely undefined and so both ;foo.bar=baz and ;bar=baz are correct expansions here. It's unclear to me if any implementations deal with hierarchical names though so I personally would never rely upon it as any two implementations could handle these differently.

To make sure this does get defined, I would submit errata.

from uritemplate-test.

To be clear, I have nothing to do with the spec

I understood that. At this point it's nice to know I wasn't just missing something obvious.

I'd argue that is completely undefined and so both ;foo.bar=baz and ;bar=baz are correct expansions here.

Agreed. Unfortunately, that makes names with . both allowed and completely unreliable/useless.

To make sure this does get defined, I would submit errata.

I was going to submit an errata if I don't manage to get a response from the authors, but I'd rather discuss it with them first in case we're missing something despite our consensus.

from uritemplate-test.

I believe your best way to get the attention of the authors is to file the errata. I don't believe they monitor this repository or issue tracker very closely

from uritemplate-test.

That's not errata material; it's something that we didn't attempt to address in the specification, as it would require defining a data model. Implementations choose how to map identifiers into data.

All that said, a new level could attempt this (as an add-on spec).

Closing this as it's not about the test suite. If you want to pursue this, a few options:

1. Send e-mail to [email protected]
2. Wait for the proposed HTTP APIs WG to start (should be very soon), and engage there
3. Write an Internet-Draft with a proposal

Cheers,

from uritemplate-test.