🇳🇱
So Gremlin Python ( Tinkerpop 3 ) does not have transaction support. So only way to get transaction like behavior is to make sure you do everything in a single database traversal.
Because I am writing an ORM / OGM model as my base that means I needed to implement something that looks like optimistic locking all my own (only works on immutable write-only graphs). Its a huge pain in the ass.
Anyway after half a day I finally figured out this is the traversal I need to resolve my optimistic lock when creating a single new node in the DB:
session.g.E().has('dirty',1).aggregate('x').fold().V().has('dirty',1).aggregate('x').choose(__.V().hasLabel('account').has("fingerprint", "DEADF00D").hasNot('dirty').count().is_(0), __.select('x').unfold().properties('dirty').drop()).iterate()
#programming #coding #Tinkerpop #tinkerpop3 #graphdb #graphtheory #python #gremlin
it looks for all nodes that have a property called "fingerprint" with a value called "DEADF00D" that does not have a property called "dirty".. If there are no nodes that match that criteria then it takes all nodes and edges with a property of "dirty" and a value of "1" and drops the dirty property from it.
I had to implement something akin to optimistic locking which means when i add a bunch of nodes and edges to the graph at first they get added one at a time with a property set called "dirty" and a value that is some sort of UUID for all the nodes/edges in the transaction.
Then I write a step that validates it and either drops the dirty property if the transaction is successful or if the transaction fails then all nodes/edges that are dirty are dropped entierly and the transaction can be reattempted.