W3C home > Mailing lists > Public > public-csv-wg@w3.org > February 2014

Re: Meeting minutes for the 19th of February...

From: Gregg Kellogg <gregg@greggkellogg.net>
Date: Fri, 21 Feb 2014 10:02:50 -0800
Cc: W3C CSV on the Web Working Group <public-csv-wg@w3.org>
Message-Id: <E38554D6-6F4E-4C9C-988C-84F6C7401608@greggkellogg.net>
To: Andy Seaborne <andy@apache.org>
On Feb 21, 2014, at 1:31 AM, Andy Seaborne <andy@apache.org> wrote:

> On 20/02/14 18:40, Gregg Kellogg wrote:
>> [[
>> andys: some json reprs e.g. gregg's could lose order
>> ]]
>> 
>> Regarding Andy's comment that JSON might loose order, that's not necessarily the case. For data in a single row, if the headers map to the same property, and we describe the container type of that property to be @list, that could keep the data in order. For example, the following context snippet:
> 
> We were going through the issues in the tabular data doc.
> 
> The discussion was "issue-1" : "Should the order of columns be significant?"  I understood that to be a general issue of columns, not just repeated names. It may be convenience/presentational. it might be a time series, e.g. "2010", "2011", "2012" case or it may be multiple groupings "region","sales", "region","sales", ....
> 
> In your example:
> 
> {
>  "name": "Markus Lanthaler",
>  "homepage": "http://www.markus-lanthaler.com/",
>  "image": "http://twitter.com/account/profile_image/markuslanthaler"
> }
> 
> how are "home", "homepage" and "image" to retain order?

Certainly, from a JSON perspective, this won't retain any value, and JSON-LD reorders keys based on lexical order in any case. I understood the discussion to be about multiple columns which relate the same value, for which we could maintain order, IMO.

> We could have metadata in the output that says "the order is (....)" but without that, any RDF representation that puts columns as property/values off a resource is going to loose the order.
> 
> RDF lists for every row would be a bad design!

Sure, but this can be specified in the meta-data on a property by property basis; for some things, it may be important, but it should not be imposed arbitrarily.

Gregg

> I'll note/add that RFC 7111 [1] uses numerical addressing so if the RDF is to relate to the URI fragments, the CSV layout needs to be (optionally) preserved.  Whether that should be a design goal, or at least making them mappable, is another question.
> 
> [1] RFC7111
> http://tools.ietf.org/search/rfc7111
> URI Fragment Identifiers for the text/csv Media Type
> M. Hausenblas, E. Wilde, J. Tennison
> 
>> 
>> {
>>   "@context": {
>>     "Value1": {"@id": "ex:value", "@container": "@list"},
>>     "Value2": {"@id": "ex:value", "@container": "@list"}
>>   }
>> }
> 
> 	Andy
> 
> 
Received on Friday, 21 February 2014 18:03:23 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:21:38 UTC