Unfortunately there are no good examples for true custom searches available. Some of this is described in the manual. Basically, you need to use a general search to pre-select some entries, for example a tags/property search with the criterion "LEVEL>=2" (this already uses the new operators, so in order to do this, make sure you have the latest devellopment version...). Then you need to write a function that will decide if the entry at point should really be included, and if not, where to continue the search. A short description of this with a small example can be found here:

On Apr 17, 2008, at 2:19 PM, Dan Davison wrote:

I'd like to ask about ways in which (sorted) sparse trees can be produced based on numerical-valued properties. One thing I have in mind is that it would seem natural to treat priority as a (1-dimensional) numerical quantity, rather than as a categorical variable as it seems to be currently. i.e. I'm attracted by the idea of being able to generate a sparse tree containing all subtrees with priority greater than 2.5, and to have headings in the sparse tree sorted by priority. (Of course I'm not suggesting replacing the current priority system; in the above by 'priority' I mean some user defined numerical-valued property.) [ Am I right in thinking that one also can't do this currently with timestamps? Not that that is necessarily needed as the agenda view does a lovely job. ]

So one version of my question would be

q1. Would it be possible to implement binary comparison operators >,<,>=,<= for use in tag and property searches? (If true numeric comparisons are difficult, then alphanumeric would still be useful.)

Sorry if this is available already.

I also have a more general question, which may well arise from my ignorance of what's available. I should admit at the outset that so far I haven't attempted to do anything other than completely basic customisations of org-mode.

q2: It's my impression that the majority of current users aren't scared of (e)lisp. (personally I am skill/knowledge-less but not scared). Is there a 'template'/'model' function a user should look at if they wanted to implement their own sparse-tree creation function? I am thinking of a function that takes as arguments a node's properties (and possibly also tags, priority, timestamps), and computes a TRUE/FALSE value (err, t/nil?) that specifies whether the subtree rooted at that point should be included in the search results. I've seen the section in the manual 'using the property API'; perhaps what I'm talking about is an extension to that API. If all users had to do were provide a predicate function include-subtree-in-sparse-tree, and org-mode deals with the tree traversal, producing the formatted output, etc, then perhaps it wouldn't be that hard for a user to implement something new like the > operators? Perhaps. Having said all that, of course I do appreciate the nice intuitive syntax for performing tag/property searches. I was just wondering about a general way of making it easy for users to extend the search functionality, and perhaps reduce the burden on the main developer of responding to requests like my q1.

Dan

_______________________________________________
Emacs-orgmode mailing list
Remember: use `Reply All' to send replies to the list.
Emacs-orgmode@gnu.org
http://lists.gnu.org/mailman/listinfo/emacs-orgmode