install server software onto `calcofiweb` per 🛣️ Roadmap

[bbest]

From README.md for Tod Kuykendall (Scripps IT) siohelp@ucsd.edu 2022-05-19 ...

🛣️ Roadmap

Marina and Renae are both UCSD staff and we are in the process of defining longer-term maintenance along with any training needed.

There are a four tiers of software ranging from most critical now (1) to ideal someday (4):

1. Database API
   The most flexible, secure way to provide the public access to the CalCOFI database is through an application programming interface (API), which can parse input arguments, execute the database query and format the results. You can visit the current prototype at api.calcofi.io (source code: plumber.R). We are currently using the R-based library Plumber (rplumber.io) to generate the API, and evaluating whether to migrate to a Python-based API generator like Flask given Marina's comfort with Python over R. A Postgresql (version 13.5) database with PostGIS spatial extension (version 3.1) is already running on the calcofiweb server that Marina is administering. The hope here is that we can also host this API on the calcofiweb server, e.g. api.calcofi.org (versus the interim instance that I am temporarily hosting at calcofi.io). See docker-compose.yml for Docker install using the rstudio service.

2. Spatial API
   The makers of the PostGIS database have created very lightweight web services with the Go programming language to provide vector tiles with pg_tileserv and GeoJSON with pg_featureserv. Try tile.calcofi.io to see the default vector tile rendering of spatial layers. These services are especially powerful APIs for developing interactive online mapping applications and reports. See docker-compose.yml for Docker install of the pg_tileserv service.

3. Apps
   The Shiny web framework makes it very easy to create applications to visualize data using htmlwidgets and responsive to user inputs and interactions. For instance, check out the app being developed by UCSB undergrads at shiny.calcofi.io/capstone. The RStudio Server provides a fully mature IDE for creating and debugging these applications, including installing required R libraries. See docker-compose.yml for Docker install of the rstudio service.

4. Portal
   Eventually, we hope to showcase how CalCOFI datasets interoperate with all the relevant portals for maximizing discovery and use across the marine oceanographic and ecological communities. By installing server node software to slice tabular and gridded datasets with ERDDAP as well as IPT for biogeographic searches, we can also highlight full metadata and all endpoints for a given dataset with CKAN. Links to the IOOS curated Docker instances and recipes for spinning these services up have been added to github.com/CalCOFI/server/issues:

• + CKAN #3
• + ERDDAP #4
• + GBIF IPT #5