Re: [Exist-open] Fwd: XQuery bug?

[Michael B. <mbe...@mb...>]

Grzegorz Chrupa=C5=82a wrote:

> Resending, it doesn't seem to have gotten through.

I think it did, but the reason there have been no responses so far may be
that your example material is pretty complex. You certainly did a helpful
thing by posting a sample document and a full query that illustrates the
problem, but people unfamiliar with your material still need to do a lot =
of
homework before they have a chance of investigating just what is wrong.

I take it that it wasn't possible for you to simplify your query without
"losing" the bug, but if that is indeed so, it is nearly always helpful t=
o
do what I suggested R=C3=A9my should do, namely test out the XPath portio=
n of
your query using the command-line client. If the XPath
//sentence//token[tags/morpho[starts-with(.,'A ')]] when run against your
collection from the command line produces no matches (or if the result se=
t
doesn't include the ones the rest of your XQuery is intended to filter) t=
hen
you have hit upon an XPath bug. If however the XPath correctly retrieves =
the
candidate nodes, then the problem lies somewhere in the XQuery processing=
.
Perhaps you could try this and report back?

Michael Beddow