Cannot access element because its enclosing transaction is closed and unbound

[Rick Richardson]

What am I doing wrong?

(def home (first (tg/transact! (tv/upsert! :roomname {:roomname "Home"
                                                :shortdesc "You find yourself in a bizarre town square"}}))))

(def users (first (tg/transact! (tv/upsert! :username {:username "Root" :type "root"}))))
(def connection (first (tg/transact! (te/upconnect! (tv/refresh users) :in (tv/refresh home)))))

clojuremud.core> connection
#<StandardEdge e[1h-8-2F0LaTPQBM][8-in->4]>
clojuremud.core> home
#<StandardVertex v[4]>
clojuremud.core> users
#<StandardVertex v[8]>

clojuremud.core> (tg/transact! (te/get-all-edges))
IllegalStateException Cannot access element because its enclosing transaction is closed and unbound  com.thinkaurelius.titan.graphdb.transaction.StandardTitanTx.getNextTx (StandardTitanTx.java:176)

clojuremud.core> (tg/transact! (tv/all-connected-vertices (tv/refresh users) :in))
IllegalStateException Cannot access element because its enclosing transaction is closed and unbound  com.thinkaurelius.titan.graphdb.transaction.StandardTitanTx.getNextTx (StandardTitanTx.java:176)

[Matt DeBoard]

FWIW I have just run into this problem as well with

          (tg/open (System/getProperty "java.io.tmpdir"))
          (tg/transact!
            (let [ex (first edges)
                  p1 (tv/create! {:name (:a ex)})
                  p2 (tv/create! {:name (:b ex)})
                  {:keys [color cost]} ex]
              (te/connect! p1 :route p2 {:color color :cost cost})))

          (tg/transact!
            (te/get-all-edges))

This is directly from the docs so I'm not sure what the deal is.

If I had to guess I'd say all the operations must be completed in a single transaction, maybe this is a by-product using the java.io.tmpdir disk persistence.

[Michael Klishin]

This issue won't be relevant with the next release which no longer relies on dynamic vars to store a graph: there will be no implicit state you may trip over between transactions (in the library).

[Matt DeBoard]

I see. I just realized you're the same guy who responded to my comment on the Neocons page. I actually started using Neocons/neo4j after this bug appeared. I went to google for it and saw a remark on the mailing list from February that Titanium was unmaintained. Is that not the case now?

[Michael Klishin]

@mattdeboard @ray1729 is working on updating it for Titan 0.4.x. This will include some long overdue public API changes. I suspect at least a few more weeks will be necessary to get Titanium, Archimedes and Orge in good shape on the new Tinkerpop stack version, so if you can't wait that long, use Neocons.

[Romain Leroux]

Any update on that ?

[Ray Miller]

The latest Titanium release (1.0.0-beta2) has the new API Michael mentioned, but the documentation is lagging behind the API changes. In the mean time, check out the tests (particularly graph_test and integration_test) for examples.

[Romain Leroux]

Thank you for your quick feedback I will check it out !