Lists should be ordered? about uritemplate-test HOT 6 CLOSED

? indicates form-style query expansion, and non-matching query parameters are unordered (i.e., the forms above are all considered equivalent).

Perhaps you're reading too much into the example variable name 'list' in the spec?

from uritemplate-test.

@mnot But what about in the / operator in that test? It permits either /person or /albums to be first.

Meanwhile, none of the examples in spec-examples.json permit reordering, e.g.:

I'm researching the reversible subset of URI Templates, and this inconsistency means I can't get both tests to pass.

from uritemplate-test.

OK, so I think this is more about the explode modifier (*), which says:

An explode modifier applied to a list variable causes the expansion to iterate over the list's member values.

... So on the face of it, the options starting with /albums/person should be removed.

@damnhandy, you added these in #5. Any thoughts? It would be interesting to see if any implementations broke if we moved these into negative-tests.json...

from uritemplate-test.

I as well expect order to be kept. Otherwise it would totally break paths, and also elements from a listbox form element whose ordering users have chosen carefully.

from uritemplate-test.

@mnot I think I agree with you here. I can probably test this out later in the week. I suspect a lot of implementations would maintain declaration order so I don't think there'd be a lot of breakage. This would need some more testing beyond a single implementation.

@awwright for years, I've had a feature request for my implementation to do reverse matching. It's issues like this one that made it more effort that it was probably worth to invest time into. The more complicated the template, the tougher the match. Had there been more specifics around ordering, it might have been easier to implement.

from uritemplate-test.

@damnhandy Yeah, as far as I can tell, the ordering is the final problem for reversing URI templates. Certain features are difficult to implement (re-using a variable increases the computational complexity from linear to exponential) or impossible (substrings). But for the most part, it's entirely possible, so long as it's deterministic (either everything needs to be an unordered set, or it needs to preserve the list ordering).

from uritemplate-test.