scheme:[//[user[:password]@]host[:port]][/path][?query][#fragment] |
|
A Data Entry is a collection of IDSs (potentially all IDSs) and their multiple occurrences (see 4.1). A Data Entry is uniquely defined by:
There is a need to have more flexibility in the pulse file location, but at the same time it’s important to keep notions of User, Database name and the new pulse_type category ("pulse", "simulation", "machine_description", "continuous_experiment", ...)). So we can’t leave it completely to an arbitrary file name and should keep some file location rules enabling keeping apart pulse files of different nature, without mixing pulse and run numbers.
Therefore the following extension/generalization is proposed : A Data Entry is uniquely defined by:
ids = imas.ids(52702, 0)
ids.open_uri("uda://g2jhollo@imas.marconi.cineca.it:56565/west/3/0")
ids = imas.ids(52702, 0)
ids.open_uri("uda://public@imas.iter.org:56565/west/3/0")
ids.open_uri("uda://imas.iter.org:56565/west/3/0")
ids = imas.ids(84600, 0)
ids.open_uri("uda://jholloc@imas.jet.uk:56565/jet/3")
The `imas.marconi.cineca.it` was just the domain I made up for where an IMAS UDA server might be running – this could be anything.
The port could be left out if default (default for UDA = 56565) but could be specified if the UDA server is running on a non-default port.
The URI scheme (uda://) would specify the backend, so `uda://…` would use the UDA backend, `mdsplus://…` would use the MDSplus backend, etc.
Currently the UDA IMAS plugin is hard-coded to using the MDSplus backend. You are right, this could be specified on the URI, perhaps as a query arg, :
ids.open_uri("uda://public@imas.iter.org:56565/west/3/0?backend=HDF5")