Describe lists containing lists, as motivated by GeoJSON. #41

gkellogg · 2018-08-05T00:45:01Z

Fixes #36.

This is implemented in w3c/json-ld-api#16.

Fixes #36.

iherman · 2018-08-05T05:27:51Z

I do not have a real issue with the changes, so we can merge it; my comments are purely of an editorial nature and reflect a matter of taste... Take them as you wish, salt 'n pepper them at taste...

I would separate the parts on sets and lists in two subsections. The two notions have a different feel and setting, and it would increase readability
The session after the table reprsenting lists in RDF you have a section that begins with "Consequently...", which is a bit out of place in this document, we usually do not analyze other serializations. I would just give the example in Turtle, and leave that (and, also, the "unwieldy" comment before the table) out. In any case, as I already raised in Explanation of examples #26, I would think we will have to rethink how we represent all our examples anyway.
You say "In JSON-LD 1.1, we can express this using recursive lists, by simply adding the appropriate context definion". This may be understood as this being the only way of doing so, which is not true. One can use repeated @list keywords for each series of coordinates, it is just unreadable...

iherman

See
#41 (comment)

gkellogg · 2018-08-05T18:51:09Z

I would separate the parts on sets and lists in two subsections. The two notions have a different feel and setting, and it would increase readability

They are both about arrays and ordering, which is why they were together. We do need to do a comprehensive reorganization, but I'd rather not do that on a piecemeal basis.

The session after the table reprsenting lists in RDF you have a section that begins with "Consequently...", which is a bit out of place in this document, we usually do not analyze other serializations. I would just give the example in Turtle, and leave that (and, also, the "unwieldy" comment before the table) out.

Noted.

In any case, as I already raised in #26, I would think we will have to rethink how we represent all our examples anyway.

Indeed, but this is a somewhat daunting task, for which I could use some help. In particular, we probably need to use something like jsonld-vis to simplify creating such graphs, although this may prove to be a problem for non-HTML use.

…sts in expanded form.

iherman · 2018-08-06T04:34:07Z

Hey @gkellogg

They are both about arrays and ordering, which is why they were together. We do need to do a comprehensive reorganization, but I'd rather not do that on a piecemeal basis.

I was not clear, sorry about that. I meant deeper subsections within this subsection. But I agree that this can wait, subject to a reorganization

Indeed, but this is a somewhat daunting task, for which I could use some help. In particular, we probably need to use something like jsonld-vis to simplify creating such graphs, although this may prove to be a problem for non-HTML use.

It is complicated indeed. The table/expanded/TriG/n-quads versions may not be that bad just lot of writing; if the graphs prove to be too much work, we can drop that one.

(B.t.w., for some reasons, example 57 does not appear in the preview. I presume it will be there for the merged version.)

azaroth42 · 2018-08-07T16:52:45Z

+1 to reorganization, and +1 to not doing it as part of this PR :)
Also ... +1 to the PR.

gkellogg · 2018-08-07T17:41:12Z

I'd like to handle this PR and w3c/json-ld-api#16, which ideally could use an implementation, but should at least have a review.

simonstey · 2018-08-07T17:48:38Z

I already pinged Axel and I'll give it a read tmrw!

simonstey · 2018-08-08T12:16:09Z

index.html

+</pre>
+
+<p class="changed">In JSON-LD 1.1, lists of lists, where the value of
+  a <a>list object</a>, may itself be a <a>list object</a> recusively, are


s/recusively// (I don't think we need to say that)

simonstey · 2018-08-08T12:17:01Z

index.html

+  a <a>list object</a>, may itself be a <a>list object</a> recusively, are
+  fully supported. For example, in <em>GeoJSON</em> (see [[RFC7946]]),
+  <em>coordinates</em> are an ordered list of <em>positions</em>, which are
+  represented as an array of two or more numbers. For example:</p>


s/numbers. For example:/numbers:/

simonstey · 2018-08-08T12:22:51Z

index.html

+}
+</pre>
+
+<p class="changed">For this example, it's important that the values


For this example,

only for this example or all uses of coordinates?

s/the values/values/

simonstey · 2018-08-08T12:23:35Z

index.html

+</pre>
+
+<p class="changed">For this example, it's important that the values
+  expressed within coordinates maintain their order, which requires the use of


s/coordinates/coordinates/ ?

simonstey

added minor comments inline

BigBlueHat · 2018-08-08T18:10:33Z

@sgillies as the primary author of GeoJSON-LD your thoughts on this would be most appreciated!

sgillies · 2018-08-11T20:48:58Z

index.html

+            [-10.0, -10.0],
+            [10.0, -10.0],
+            [10.0, 10.0],
+            [-10.0, -10.0]


I feel that GeoJSON examples can be made less confusing with respect to longitude, latitude ordering by using real-world coordinates like [-105.615115, 40.255014] (latitude: 40.255 degrees N, longitude: 105.615 degrees W).

Note that the lists of lists feature does not preclude #34, although it becomes less important, and it does add significantly more complexity to the processing model.

sgillies · 2018-08-11T20:51:41Z

index.html

+    "@vocab": "https://purl.org/geojson/vocab#",
+    "type": "@type",
+    "bbox": {"@container": "@list"},
+    "coordinates": {"@container": "@list"}


In geojson/geojson-ld#34 we discussed an @ndarray, but @list looks just as good, or better in that it's easier to explain and understand.

sgillies · 2018-08-11T20:53:58Z

index.html

+          {"@list": [{"@value": -10.0}, {"@value": -10.0}]},
+          {"@list": [{"@value": 10.0}, {"@value": -10.0}]},
+          {"@list": [{"@value": 10.0}, {"@value": 10.0}]},
+          {"@list": [{"@value": -10.0}, {"@value": -10.0}]}


This is fine. Tagential question: is there a way to indicate to JSON-LD users that the first element of these 2-tuples is longitude and the second is latitude?

Not really. If the vocabulary were modeled in OWL, there may be some way to do this at that level, @iherman may know better.

To be honest, even in OWL I don't see a way to do that.

@sgillies FWIW, we've got an issue to discuss/address that. See: #7 "support @values for describing multidimensional containers"

From a graph modeling perspective, the fact that the lat/long were originally presented in a list isn't important, is it? Only the "outer" list of coordinates has value to the graph (afaict).

What @gkellogg and @iherman said. In JSON-LD it will just be surface syntax of arrays. At the RDF level, it will be nested rdf:Lists with no further semantics. If more semantics are required, then a different structure would also be required.

Describe lists containing lists, as motivated by GeoJSON.

618258b

Fixes #36.

gkellogg requested review from iherman and simonstey August 5, 2018 00:45

Fix turtle syntax.

30adecd

iherman approved these changes Aug 5, 2018

View reviewed changes

Factor in @iherman's comments, and include an example of recursive li…

31d1804

…sts in expanded form.

azaroth42 self-requested a review August 7, 2018 16:52

azaroth42 approved these changes Aug 7, 2018

View reviewed changes

simonstey reviewed Aug 8, 2018

View reviewed changes

simonstey approved these changes Aug 8, 2018

View reviewed changes

sgillies mentioned this pull request Aug 11, 2018

Develop a @ndarray container type geojson/geojson-ld#34

Closed

sgillies reviewed Aug 11, 2018

View reviewed changes

Address comments from @simonstey.

b754d86

gkellogg merged commit 40d18d5 into master Aug 11, 2018

gkellogg deleted the issue-36-lists-of-lists branch August 11, 2018 21:49

BigBlueHat mentioned this pull request Aug 14, 2018

Feature Detection in JSON-LD Processors #33

Closed

almereyda mentioned this pull request Feb 8, 2019

Spatiotemporal mapping valueflows/forum.valueflo.ws#27

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Describe lists containing lists, as motivated by GeoJSON. #41

Describe lists containing lists, as motivated by GeoJSON. #41

gkellogg commented Aug 5, 2018 •

edited by pr-preview bot

Loading

iherman commented Aug 5, 2018

iherman left a comment

gkellogg commented Aug 5, 2018

iherman commented Aug 6, 2018

azaroth42 commented Aug 7, 2018

gkellogg commented Aug 7, 2018

simonstey commented Aug 7, 2018

simonstey Aug 8, 2018

simonstey Aug 8, 2018

simonstey Aug 8, 2018

simonstey Aug 8, 2018

simonstey left a comment

BigBlueHat commented Aug 8, 2018

sgillies Aug 11, 2018

gkellogg Aug 11, 2018

sgillies Aug 11, 2018

sgillies Aug 11, 2018

gkellogg Aug 11, 2018

iherman Aug 12, 2018

BigBlueHat Aug 14, 2018

BigBlueHat Aug 14, 2018

azaroth42 Aug 14, 2018

Describe lists containing lists, as motivated by GeoJSON. #41

Describe lists containing lists, as motivated by GeoJSON. #41

Conversation

gkellogg commented Aug 5, 2018 • edited by pr-preview bot Loading

iherman commented Aug 5, 2018

iherman left a comment

Choose a reason for hiding this comment

gkellogg commented Aug 5, 2018

iherman commented Aug 6, 2018

azaroth42 commented Aug 7, 2018

gkellogg commented Aug 7, 2018

simonstey commented Aug 7, 2018

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

simonstey left a comment

Choose a reason for hiding this comment

BigBlueHat commented Aug 8, 2018

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

gkellogg commented Aug 5, 2018 •

edited by pr-preview bot

Loading