Alain,
I disagree that "a tree is a tree" in the manner that you seem to mean it.
There are very significant differences between XML and JSON, and nor
merely in their surface syntax. The most obvious such difference is
that XML trees can be "annotated" (with attributes, namespaces, et al),
while JSON trees cannot be. Another is that JSON trees can incorporate
a node type completely foreign to XML trees -- arrays.
With tongue very slightly in cheek, I'll point out that the X in
XPath/XQuery stands for "XML", and not for "Tree" or "JSON" or anything
else. We could have created a JPath/JQuery, or maybe a TPath or TQuery,
but our remit was to product a path language and a query language for
XML. Having JSON support in that same language was not exactly trivial
or uncontroversial, but we knew/know that JSON is a very important data
model these days.
Hope this helps,
Jim
On 4/10/2016 11:36 PM, Alain Couthures wrote:
> Well, a tree is a tree, no matter if its notation is this or that...
>
> It is a key point for XQuery/XPath to support axes. In my own dev
> context, I would not say that I use ancestors or siblings axes often
> but they really saved my days each time and with a clean and concise
> syntax. As an implementor, I know that this has an effective cost in
> memory and speed and that it might be tricky to optimize.
>
> I understand that performance is important but there are many
> situations where data to be processed is not that big or where batch
> processing can be performed on underemployed servers.
>
> Treating JSON more lightly than XML is a little bit strange for me:
> why not, for the same "bad" reasons, decide to manage no axes except
> descendants for XML too??
>
> All this might just be viewed as a runtime optimization for
> implementations, don't you agree?
>
> Regards,
>
> Alain Couthures
>
> Le 11/04/2016 06:09, Dimitre Novatchev a écrit :
>>> This apparently trivial difference has some interesting and profound
>>> consequences.
>>> Transformations of a tree in which many branches of the tree remain
>>> unchanged become much more efficient,
>>> because (in an environment where in-situ update is not allowed)
>>> copying a subtree has zero cost.
>> To give a specific and simple example
>>
>> 1. Sharing a subtree (between two trees) in a binary search tree has
>> time complexity just O(log(N)) -- the time needed to find the node in
>> a BST of N nodes. Compare to having to create a complete copy of the
>> tree (which is necessary if a node can have no more than one parent)
>> -- O(N). The same for creating a tree to which just one node has been
>> changed (replaced) -- all tree nodes can be shared with the exception
>> of the nodes on the path from the original root to the node that must
>> be changed.
>>
>> 2. Sharing a subtree between M trees (or creating a tree from the
>> original one in which M nodes have to be changed) -- O(M(logN)).
>> Compare this to creating M complete copies of the tree (which is
>> necessary if a node can have no more than one parent) -- O(M*N).
>>
>> To put it simply, the difference in performance is similar to that
>> between Quicksort and BubbleSort: O(N*log(N)) / O(N^2)
>>
>> The same for space complexity.
>>
>>
>> Cheers,
>> Dimitre
>>
>>
>> On Sun, Apr 10, 2016 at 2:20 PM, Michael Kay <
mike@saxonica.com> wrote:
>>>> I was sorry to read that XQuery 3.1 is not considering JSON data as
>>>> a tree: JSON is treated as a "complex" atomic type, from my point
>>>> of view.
>>>>
>>> You can regard the output of parse-json() as a tree (of maps,
>>> arrays, etc), but it is different from the tree representation of
>>> XML. The main difference is that the maps, arrays, etc produced by
>>> parse-json() have no parent.
>>>
>>> This corresponds closely to the way JSON data is handled in other
>>> programming languages.
>>>
>>> This apparently trivial difference has some interesting and profound
>>> consequences. Transformations of a tree in which many branches of
>>> the tree remain unchanged become much more efficient, because (in an
>>> environment where in-situ update is not allowed) copying a subtree
>>> has zero cost. However, the lack of upwards or sideways navigation
>>> means that during a recursive descent of the tree, you need to pass
>>> a lot more context.
>>>
>>> For an exploration of the effect of these differences on some
>>> transformation use cases, see my XML Prague 2016 paper.
>>>
>>> Michael Kay
>>> Saxonica
>>>
>>>
>>>
>>> _______________________________________________________________________
>>>
>>> XML-DEV is a publicly archived, unmoderated list hosted by OASIS
>>> to support XML implementation and development. To minimize
>>> spam in the archives, you must subscribe before posting.
>>>
>>> [Un]Subscribe/change address:
http://www.oasis-open.org/mlmanage/>>> Or unsubscribe:
xml-dev-unsubscribe@lists.xml.org>>> subscribe:
xml-dev-subscribe@lists.xml.org>>> List archive:
http://lists.xml.org/archives/xml-dev/>>> List Guidelines:
http://www.oasis-open.org/maillists/guidelines.php>
--
========================================================================
Jim Melton --- Editor of ISO/IEC 9075-* (SQL) Phone: +1.801.942.0144
Chair, ISO/IEC JTC1/SC32 and W3C XML Query WG Fax : +1.801.942.3345
Oracle Corporation Oracle Email: jim dot melton at oracle dot com
1930 Viscounti Drive Alternate email: jim dot melton at acm dot org
Sandy, UT 84093-1063 USA Personal email: SheltieJim at xmission dot com
========================================================================
= Facts are facts. But any opinions expressed are the opinions =
= only of myself and may or may not reflect the opinions of anybody =
= else with whom I may or may not have discussed the issues at hand. =
========================================================================
