pure xquery implementation about xq-promise HOT 6 OPEN

James, you need to get a twitter account ;-)

from xq-promise.

Weird how things work. I literally realized this myself last night and have begun work.

See my comment on #12

The pattern could be completely realized with closures. John from Marklogic eluded to this idea at the conference but it didn't click at the time.

Cheers!

from xq-promise.

declare function local:defer($work as function(*), $args as item()*) {
  function () {
    fn:apply($work, array { $args })
  }
};

let $greet := function ($name) { 'Hello ' || $name }
let $promise := local:defer($greet, 'world')
return
  $promise()

from xq-promise.

In Marklogic, you could implement fork immediately with a pure XQuery 3.1 implementation simply with your spawn-function. Then fork-join could be implemented with pure XQuery using the new fork function.

Cheers!

from xq-promise.

All tests pass except transform clause which requires the fix from Christian which he just merged yesterday. Checkout the xq-promise.xqm file for the identical implementation with pure XQuery.

(Had to remove fork and only support fork-join) for now.

from xq-promise.

James; thanks for the new version, which really looks great, and which you implemented in hardly any time! I looked at the sources, and spontanteously, I have nothing to add that’s really noteworthy. Let’s stlil try, though…

It sounds a bit strict, but I think it could make sense to only accept arrays as function arguments (for p:invoke and p:defer). This it might reduce confusion if actual arrays are passed on as arguments:

p:defer(function($a) { $a instance of xs:integer }, 1),  (: true :)
p:defer(function($a) { $a instance of array(*)   }, [1]) (: false?... :)

(: ...or... :)

p:defer( function($a as array(*)) { (: ... :) }, [ 1,2 ] ),    (: wrong :)
p:defer( function($a as array(*)) { (: ... :) }, [ [ 1,2 ] ] ) (: correct :)

from xq-promise.