Re: [Json] [Cbor] JSON and int64s - any change in current best practice since I-JSON

[Jeremy O'Donoghue <[e-mail]>]

Hi Carsten,

Proposed text looks good to me.

I like test vectors, but I agree that they represent a lot of work. I won’t die on that hill or I might end up with the action to create them 😊

Jeremy

On 29/01/2024, 10:05, "Carsten Bormann" <[e-mail]> wrote:

WARNING: This email originated from outside of Qualcomm. Please be wary of any links or attachments, and do not enable macros.

On 2024-01-29, at 10:54, Jeremy O'Donoghue <[e-mail]> wrote:
>
> “Note: Implementors are advised that the built-in numeric parsers in some programming implementations accept only a subset of the floating point syntax defined in [IEEE 754] and used in this document. In particular, care may be required with respect to supporting input with leading and/or trailing zeros.”

Thanks!

I combined this with the text already there, streamlined it a bit, and came up with:

https://github.com/cbor-wg/edn-literal/pull/30

Please have a look!

> Might also be good to have a couple of examples which are known to fail with at least some language parsers. At least for myself, the first thing I do in an implementation is to implement all of the examples in an RFC as test cases, so this is an easy way to ensure that implementers catch any issues.

There are some examples in that note; I’m not sure we should do a full-on attempt at defining test vectors in this document though.

Grüße, Carsten