Re: [xml-dev] XQuery Adoption Survey Results

[Michael Champion <michaelc.champion@gmail.com>]

On Tue, 29 Mar 2005 11:34:39 -0500, Tony Lavinio <xml1@lavinio.net> wrote:
> Obviously, XQuery vendor adoption is huge, but what about
> developer adoption? Are people actually using XQuery?
> 
> DataDirect conducted a survey of 550 software professionals and
> found that 52% of those surveyed were already using XQuery, and
> that another 33% said that they plan to use XQuery in the next
> 12 months -- Incredible considering the official standard from
> the W3C is still several months from final completion. The free
> study also investigates who is using XQuery, and why. You can
> download the results from here:
> http://www.datadirect.com/xquerystudy/

The whole point of a survey reporting percentages and margins of error
is to generalize about some larger population of people.  What is the
population from which the sample was drawn?  If it is a representative
sample of the overall population of software professionals, these
results are indeed incredible ... perhaps in the literal sense of the
term :-).  If it is 550 people selected from the overall population of
XML developers, the results would still be quite intriguing; I would
be very surprised to find this much interest in XQuery given the
general lack of XQuery enthusiasts on this list.  But the list isn't
exactly a representative sample of XML developers.

If this is 550 people selected from the overall population of people
who build XML-centric applications on top of a DBMS, the numbers sound
about right to me.  That would confirm my understanding that XQuery is
the obvious choice for an XML DB query language and that all other
contenders have dropped off the radar. Likewise, since XQuery
implementations are starting to mature, this survey could be valuable
evidence that a "meme" is starting to proliferate of using XQuery on
top of a database to handle large volumes of XML rather than streaming
it thru SAX or whatever.

But if  it is 550 people who happened to be your customers or were
selected from some group known to have a high percentage of XQuery
users, the results are more or less meaningless to the rest of us. 
Even M. David Peterson  probably admits that there are a few hundred
XQuery users in the world :-)  The question is whether you just found
them and gave them your survey, or if those 550 are representative of
a much larger and more interesting population.

So, in statistical terms, who are these developers who are rapidly
adopting XQuery?

p.s. The most intriguing bit of the survey results for me was the
aside "DataDirect's own informal tests, which have shown that for
various common use cases involving querying and transforming XML data
sources, an XQuery expression can accomplish the desired result using
just 1/5th of the code of acomparable XSLT stylesheet solution; and
just 1/20th of the coderequired by a lower-level DOM programming
approach."  I bet a lot of us would like to see those common use cases
and judge for ourselves whether [D|X|JD]OM+XPath, XSLT, or XQuery
differ dramatically in their effectiveness for addressing them.   My
personal guess is that in general XPath does the heavy lifting in all
three approaches and whether one approach is more efficient to write
than another depends largely on the problem to be solved (maybe XQuery
is easier for schema-valid XML data, XSLT for deeply recursive
documents, an API for situations where a lot of ad hoc function calls,
calculations, and transformations are needed).  If you can refute
that, I would be VERY interested.