






ES3 notes
Meeting 05/23/2005
Attendees: Frew, Peter
ES3 Design overview
- ES3 'core' is designed so that the client must tell it the links between objects (i.e. lineage info)
- the probulator collects sufficient info so that ES3 can determine these links
- the probulator only generates metadata from concrete objects: files and transformations (scripts, .pro files, etc.)
- Peter postulates that 'abstract' metadata is difficult to comprehend
- 'abstract' metadata means:
- the metadata may not actually refer to a concrete object
- OR the user manually makes the connection between the metadata and a concrete object it describes
- ES3 should focus on metadata that refers to concrete objects
- users will understand more clearly what ES3 does
- user interface (client library) will be clearer, cleaner
- ES3 more dependant on probulator
- ES3 client library will still support 'abstract' metadata but user will have to manually create lineage info
- annotation of ES3 objects
- ES3 must support asynchronous annotation of ES3 objects
- e.g. Tom Painter may wish to describe scientific model that ES3 object 'refl_retrieval' refers to
- this annotation refers to a set of instances of 'refl_retrieval'
- annotation can be sent to ES3 before, during or after the object(s) it describes are sent to ES3
- user specifies the set of objects the annotation refers to by using a 'select' statement or similiar type of qualification that will be stored in ES3
Link to this Page
- Working Documents last edited on 8 February 2008 at 8:58 am by pool-71-102-172-126.snloca.dsl-w.verizon.net