gistfile1.clj

(use 'forma.hadoop.jobs.scatter)

;; put stuff in pail
(let [pail-path "/Users/robin/delete/tmp/newpail"
      vcf-src [(thrift/DataChunk* "vcf"
                                  (thrift/ModisPixelLocation* "500" 28 8 0 0)
                                  25
                                  "00")]
      gadm-src [(thrift/DataChunk* "gadm"
                                   (thrift/ModisPixelLocation* "500" 28 8 0 0)
                                   16435
                                   "00")]
           vcf-query (<- [?dc]
                         (vcf-src ?dc))
           gadm-query (<- [?dc]
                          (gadm-src ?dc))]
  ;; Data appear in pail no problem:
  (to-pail pail-path vcf-query) 
  (to-pail pail-path gadm-query))



(defn unpack-long
  "Unpack long value from static source"
  [dv]
  (.getLongVal dv))

;; get values out of pail
;; data comes out when there's one source, but joining on sres h v s l produces nothing

;; This works:
(let [pail-path "/Users/robin/delete/tmp/newpail"
      vcf-src (split-chunk-tap pail-path ["vcf"])]
  (<- [?sres ?h ?v ?s ?l ?vcf]
      (vcf-src _ ?vcf-obj)
      (thrift/unpack ?vcf-obj :> _ ?vcf-loc ?vcf-dv _ _)
      (thrift/unpack ?vcf-loc :> ?sres ?h ?v ?s ?l)
      (unpack-long ?vcf-dv :> ?vcf)))

;; Works but no results:
(let [pail-path "/Users/robin/delete/tmp/newpail"
      vcf-src (split-chunk-tap pail-path ["vcf"])
      gadm-src (split-chunk-tap pail-path ["gadm"])
      ]
  (<- [?sres ?h ?v ?s ?l ?vcf ?gadm]
      (vcf-src _ ?vcf-obj)
      (thrift/unpack ?vcf-obj :> _ ?vcf-loc ?vcf-dv _ _)
      (thrift/unpack ?vcf-loc :> ?sres ?h ?v ?s ?l)
      (unpack-long ?vcf-dv :> ?vcf)

      (gadm-src _ ?gadm-obj)
      (thrift/unpack ?gadm-obj :> _ ?gadm-loc ?gadm-dv _ _)
      (thrift/unpack ?gadm-loc :> ?sres ?h ?v ?s ?l)
      (unpack-long ?gadm-dv :> ?gadm)))