search

End-to-end-provenance / DDG-Explorer

Software to support creation, visualization and querying of Data Derivation Graphs.
5 stars 5 forks source link

Query results stop at function boundaries #13

Open blernermhc opened 9 years ago

blernermhc commented 9 years ago

The Derived From and Results queries follow data flow edges to trace back how data is computed or what is computed from it. With the current recording of information for function calls and returns, the data flow path is incomplete because we do not record those details within the function. The result is that we see the return statement that returned a value, but not what parameters were passed to the function to help it do the computation.

To solve this, it might be best to store more information in the database. I'm not sure exactly what we need but something that lets us easily go from a Finish node for a function to finding out what parameters were passed to that function. Maybe storing the collapsed nodes that DDG Explorer creates along with the edges going into and out of them would solve the problem.

blernermhc commented 8 years ago

It now goes through the function calls. However, the layout is really bad! See, for example, DailySolarRadiation and do queries about how input files are used and output files are produced.

blernermhc commented 7 years ago

It might be improved if we sort the procedure nodes by their ids before adding them to the Prefuse graph.