A couple of possible changes for the next ICAT release

[Steve Fisher]

I would like to make a couple of changes for the next ICAT release but want to discuss them here before adding to the list of requirements.

Firstly I would like to remove (or at least disable) the notification mechanism. I have a number of reasons for this: 1) The security model is not clean, 2) is does not meet current needs to understand how ICAT is being used (for which I think a log file analyser is probably better suited)  and 3) it slows ICAT down and adds complexity. It may be possible to bring it back later in a slightly different form. Is anybody currently using it please and would anybody like to use it in the future.

I would also like to simplify the schema by removing nullable components from unique keys. In some cases the column should be made NOT NULL and in other cases it should be removed from the key. For example:

For Dataset I would remove both type and sample so that a dataset is identified just by its name and investigation

For Datafile this should be identified by its name and dataset

For ParameterType I would like it to be identified by Facility and name - but not units. I suspect that people won't agree to this however.

These simplifications would increase database portability,  would probably increase performance and simplify the design of the icat data service.

Steve

[tom.g...@stfc.ac.uk]

Hi Steve,

 

> Firstly I would like to remove (or at least disable) the notification mechanism.

 

I agree with 1 and 3. Slightly remain to be convinced on 2. I have concerns about the longevity of logs. They often get ‘archived’ or are not moved between servers. Maybe using the log4J database appender would resolve this fear.

My other use case for notifications is automated analysis – trigger some Mantid job when a datafile is catalogued. I could do this as part of my ingest process.

 

> For Dataset I would remove both type and sample so that a dataset is identified just by its name and investigation

 

I am happy with this

 

> For Datafile this should be identified by its name and dataset

 

Agree

 

> For ParameterType I would like it to be identified by Facility and name - but not units. I suspect that people won't agree to this however.

 

Not keen on this - doesn’t work for me unless I include units in the name (ugly), or do some conversion on ingest –I don’t always know how to do this as parameter names and units are configurable by the instrument scientist or user : they change a lot and without warning.

 

Tom

--
http://groups.google.com/group/icat-developers

--
Scanned by iCritical.