No, never thought of ** instead of *. Also not in my printout of the manual, but that might be fairly outdated.
Thanks.

By the way, did you see my post on \xmllastatt from yesterday?

On 12 Sep 2018, at 13:23, Hans Hagen <j.hagen@xs4all.nl> wrote:

On 9/12/2018 1:03 PM, Hans van der Meer wrote:
There is a difference of behaviour in \xmlfilter between
(1) \xmlfilter{#1}{node/command(xmlsetup)} and
(2) \xmlfilter{#1}{*/command(xmlsetup)}
(3) \xmlfilter{#1}{*/*/command(xmlsetup)}
In case (1) the lpath ‘node’ results in processing of the <wildcard> nodes in all child nodes.
In case (2) and (3) no more than the corresponding level is searched.
Thus there is a difference in behaviour between node and * as an lpath.
Very unfortunate for me, because I need to process all nodes regardless of theeir node name, but selected on the basis of an attribute: \xmlfilter{#1}{*[@attribute==“yes”].
The difference in behaviour makes this impossible, because a general proceudre is impossible and doing a search with * and */* and */*/* etc. always will be one level too short, I expect.
Is this intended behaviour, and if so, is there a general solution?

did you try **

-----------------------------------------------------------------
                                         Hans Hagen | PRAGMA ADE
             Ridderstraat 27 | 8061 GH Hasselt | The Netherlands
      tel: 038 477 53 69 | www.pragma-ade.nl | www.pragma-pod.nl
-----------------------------------------------------------------

dr. Hans van der Meer