Jacek,

Thanks for the comments. The split you suggested is already in the schema,
e.g. Pipeline + PipelineRun, ditto for ProcStep and Component. The idea is
the same, one encodes "static" provenance, the other info relevant to
runtime. I can rename the tables to agree with UML.

Thanks for the info on Data Product. I think our whole database is a "data
product", so perhaps this provenance is captured outside the database
(e.g. version of the data release)?

Thanks,
Sergei.


> Hi Sergei,
>
> Great. See comments below.
>
>
>> CREATE TABLE Pipeline (
>>     -- Provenance data for pipelines.  Placeholder.
>>     --
>>     pipelineId  INT NOT NULL UNIQUE,
>>     -- Id for the pipeline
>>     pipelineVersion
>>                 INT,
>>     -- Version of the pipeline
>>     name        VARCHAR(255),
>>     -- Name of the pipeline
>>     author      VARCHAR(255),
>>     -- Name of developer
>>     comments    VARCHAR(255)
>> );
>
> Have a look at the UML:
> Logical View
>    Domain Model
>      Framework
>        <<entity>> Pipeline Provenance
>
>
> We should resync your table with what is in uml.
>
> I suggest we split your table into two tables:
>
> CREATE TABLE Pipeline (
>    -- contains static info about pipeline
>    pipelineId  INT NOT NULL UNIQUE, -- Id
>    name        VARCHAR(255),  -- Name of the pipeline
>    author      VARCHAR(255),  -- Name of developer
>    comments    VARCHAR(255)
> );
>
> and
>
> CREATE TABLE PipelineProvenance (
>    pipelineId  INT NOT NULL UNIQUE, -- link to Pipeline
>    pipelineVersion  INT,      -- Version
>    startTime        DATETIME  -- (see uml)
>    stopTime         DATETIME  -- (see uml)
>    processingSteps  ...       -- (see uml)
>    parameters       ...       -- (see uml)
> );
>
>
>
>> CREATE TABLE ProcStep (
>>     -- Provenance data for processing steps.  Placeholder.
>>     --
>>     procStepId  INT NOT NULL UNIQUE,
>>     -- Id for the processing step
>>     procStepVersion
>>                 INT,
>>     -- Version of the processing step
>>     name        VARCHAR(255),
>>     -- Name of the proc. step
>>     author      VARCHAR(255),
>>     -- Name of developer
>>     comments    VARCHAR(255)
>> );
>
>
> same comments as for Pipeline
>
>
>> CREATE TABLE Component (
>>     -- Provenance data for components.  Placeholder.
>>     --
>>     componentId INT NOT NULL UNIQUE,
>>     -- Id for the component
>>     componentVersion
>>                 INT,
>>     -- Version of the component
>>     name        VARCHAR(255),
>>     -- Name of the component
>>     author      VARCHAR(255),
>>     -- Name of developer
>>     comments    VARCHAR(255)
>> );
>
>
>
> same comments as for Pipeline
>
>
> BTW, the 4th provenance that is in uml is for
> Data Product. It would probably be useful to add
> it to our schema as well.
>
>
> Thanks,
> Jacek
> _______________________________________________
> LSST-data mailing list
> [email protected]
> http://www.lsstmail.org/mailman/listinfo/lsst-data
>

_______________________________________________
LSST-data mailing list
[email protected]
http://www.lsstmail.org/mailman/listinfo/lsst-data

Reply via email to