W3C home > Mailing lists > Public > public-linked-json@w3.org > July 2013

RE: API edits to avoid implying that JSON-LD is not RDF

From: Markus Lanthaler <markus.lanthaler@gmx.net>
Date: Tue, 30 Jul 2013 18:40:49 +0200
To: "'Linked JSON'" <public-linked-json@w3.org>
Message-ID: <00aa01ce8d43$8cf6c1b0$a6e44510$@lanthaler@gmx.net>
Hi David, all,

I've converted your changes to the API spec to a pull request and commented
your changes there:


I could live with most changes even though I think they make the document
more difficult to read. For example, would you be OK with changing the
algorithm names slightly? I would prefer to call them

  Deserialize JSON-LD to RDF Algorithm
    Instead of Deserialize to RDF Algorithm

  Serialize RDF to/as JSON-LD Algorithm
    instead of Serialize from RDF Algorithm

to make it clearer what they are doing.

The other comments can be found directly in the PR at


I think it's easier to discuss them (in the proper context) there.


Markus Lanthaler

> -----Original Message-----
> From: David Booth [mailto:david@dbooth.org]
> Sent: Monday, July 29, 2013 4:59 PM
> To: Markus Lanthaler
> Cc: 'Linked JSON'
> Subject: Re: API edits to avoid implying that JSON-LD is not RDF
> On 07/29/2013 10:29 AM, Markus Lanthaler wrote:
> > On Monday, July 29, 2013 4:03 PM, David Booth wrote:
> >> On 07/28/2013 11:01 PM, Melvin Carvalho wrote:
> >> [ . . . ]
> >>> There's slightly more than serialization / deserialization going on
> >>> IMHO.  For example the JSON Number is coerced to XSD integer /
> double
> >>> which are not 100%  the same things, depending on the
> implementation.
> >>>
> >>> I dont think these changes are terrible, but unless I've missed
> >>> something, convert seems to be accurate here, and I'd lean towards
> >>> keeping things the same.
> >>
> >> Yes, and I actually kept the word "convert" when discussing lower-
> level
> >> things like data type conversions, for exactly the reason that you
> cite.
> >>    The use of phrases like "serialize from RDF" and "deserialize to
> RDF"
> >> is only at the higher level, in discussing the overall process.
> >
> > I haven't looked at the changes yet but I tend to agree with Melvin.
> These
> > are not serialization algorithms, i.e., they are not producing an
> output in
> > any of the concrete RDF syntaxes but emit abstract data structures
> > representing RDF's data model, i.e., triples/quads. So, what if we
> would
> > call those algorithms "Convert to/from RDF Quads Algorithm" instead?
> No, the serialization is in the other direction: from abstract RDF to
> concrete JSON-LD.  Going from concrete JSON-LD to abstract RDF is
> deserialization.
> >
> > To be honest, I don't really see why we can't keep the current names
> given
> > that the algorithms convert abstract RDF to a concrete JSON-LD
> serialization
> > or vice versa.
> In that sentence, you just described the overall process as
> serialization, and the lower-level techniques as conversion, and that
> is
> *exactly* the terminology strategy that I used, though it is possible
> that a drew the line in a slightly different place than you would.
> Bear
> in mind that the purpose of the edits is to avoid implying that JSON-LD
> is not RDF.  Phrases like "convert JSON-LD to RDF" wrongly imply that
> JSON-LD is not RDF.  That's why I am trying to use the terms
> "serialize"
> and "deserialize" when talking about the overall process, while using
> terms like "convert" and "transform" when talking about specific
> actions
> in the algorithms.
> Please read through the actual changes and see if you still have
> concerns.
> David
Received on Tuesday, 30 July 2013 16:41:19 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 16:18:38 UTC