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
_______________________________________________________________________
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
