This week I (re-)read the Anti-Patterns book, which I first read more than 10 years ago. The idea of Anti-Patterns is that there are various unproductive behavior modes in software engineering which result in similar problems, and can be treated in similar ways. This book is well worth a review now and again.

 I think the anti-pattern most relevant to the netCDF libraries is Project Mismanagement, for which I blame no one other than myself. ;-0

It's interesting that the solution to this anti-pattern, risk analysis, is something that is already underway for future netCDF releases.

To anyone tired of waiting for the configure script to run: try the -C option.

This creates a configure cache (config.cache) which stores the result of each test. When a test is to be run, configure first checks the cache, and if the answer is there, the test is not run. This helps a lot when rerunning configure, but also running it just one time. This is because configure reruns a lot of tests multiple times.

This also speeds up the libtool and udunits configure runs, which will use the answers from the netcdf configure run if they are available.

When model data is encoded in GRIB, the dataset schema is lost. So is an unambiguous identification of the the schema variables. To deal with these problems, GRIB reading libraries probably make assumptions that may not be valid for all datasets.

Problems with external tables make BUFR/GRIB not suitable as long-term storage formats. To solve this, there must be a foolproof way for reading software to know what tables the writing software used, and there must be an authoritative registry of tables.

Here's a great demo, less than 5 minutes long, of Tools UI and using NcML: https://www.youtube.com/watch?v=w-Df4Np7tLA