Best practices - feature datasets and capitalization for File Geodatabases

Question

Hello,

I have a single File Geodatabase that contains data in two projections.  The database itself has 14 feature classes of which seven are in one projection and seven are in another projection.

Question 1:

Should I keep all the feature classes in the "root" folder of the Geodatabase, or should I have two separate feature datasets for each projection?

Question 2:

Should my File Geodatabase, Feature Dataset, and Feature Classes use proper case or lower case?

Thanks

LyonMNGIS · Answer

Thank you everyone for your responses.  I also reviewed some of the links and it seems to me that there are set standards, but a lot of this is more of a personal preference.  So here is what I am going to do but please let me know if I am wrong.

* snake_case is better if we expect our data to be loaded into an RDBMS.  ProperCase is okay.
* The use of Feature Datasets is a matter of preference for GeoDatabases with a handful of layers, recommended when a Geodatabase has many layers, and required for complex datasets.
* Singular is better than plural.  - But this is going to be very hard for myself to change 🙂
* Never use spaces, punctuation, or SQL statements.
* Always use YYYYMMDD for dates.
* Describe a relationship such as ParcelHasOwners.

Thank you again for your thoughts!

VinceAngelo · Answer

Case is a bit tricky -- Geodatabases are case-insensitive, but the enterprise geodatabase tables themselves require single-case (UPPER for Oracle, lower for others). ArcGIS clients will show the mixed-case alias for the tables, but Esri doesn't support the RDBMS extensions that allow actual mixed case (quotes around names, e.g., "MixedName").

I guess there is merit in using mixed case in aliases, since it will subtly allow you to detect if the table is registered.

- V

MobiusSnake · Answer

I never use feature datasets unless they're necessary (complex dataset types like topologies, UNs, and so on).  Feature datasets mean shared locks, and although that generally has less of an impact in FGDBs than in EGDBs, it's just an unnecessary complication for what I consider minimal benefit.