RE: integer in EXI4JSON

Hi Daniel, John and all,

According to JSON specification that you provided a link for, IEEE754 seems to
be just a practice suggested for interoperability. I believe some financial applications
such as XBRL heavily depend on xsd:decimal. Maybe we should stop short of
removing decimal as John is suggesting.

Thank you,

Takuki Kamiya
Fujitsu Laboratories of America


From: Peintner, Daniel [mailto:daniel.peintner.ext@siemens.com]
Sent: Tuesday, March 28, 2017 12:09 AM
To: John Cowan <cowan@ccil.org>; Takuki Kamiya <tkamiya@us.fujitsu.com>
Cc: public-exi@w3.org
Subject: AW: integer in EXI4JSON


Hi John,

Thanks for your feedback.

So far there was no-one strongly arguing to keep "decimal". Good to hear that there seems some rationale.

Having said that, I was looking at the IETF draft for JSON number [1] and it seems best practice to remain in IEEE 754 range which we do with xsd:double.

Have you seen many use-cases in the wild going beyond these ranges?

Thanks,

-- Daniel

[1] https://tools.ietf.org/html/rfc7159#section-6



________________________________
Von: John Cowan [cowan@ccil.org]
Gesendet: Dienstag, 28. März 2017 03:55
An: Takuki Kamiya
Cc: Peintner, Daniel (ext) (CT RDA NEC EMB-DE); public-exi@w3.org<mailto:public-exi@w3.org>
Betreff: Re: integer in EXI4JSON

On Mon, Mar 27, 2017 at 8:50 PM, Takuki Kamiya <tkamiya@us.fujitsu.com<mailto:tkamiya@us.fujitsu.com>> wrote:

In the case of "decimal", probably we have not seen a good use case to use it in EXI4JSON.
I tend to agree that we drop "decimal".

If you leave it out, it is not possible to do 1:1 conversion between arbitrary JSON and EXI4JSON.  For example, a number greater than 1e308 can only be represented as decimal, or a number that requires more than 17 digits of precision.

--
John Cowan          http://vrici.lojban.org/~cowan        cowan@ccil.org<mailto:cowan@ccil.org>
LEAR: Dost thou call me fool, boy?
FOOL: All thy other titles thou hast given away:
That thou wast born with.