Running the bootstrap procedure, when and how

[arni-magnusson]

@jimianelli raised a good point, accompanied with a pull request #2, asking whether the data.R script should start with a call to taf.bootstrap().

One thing that's important - but not quite documented yet - is that the bootstrap procedure should not be run every time the analysis is run.

This is because in many cases, the bootstrap procedure fetches data from databases. We want to be able to always run with the same data, even when the underlying databases keep changing. So things inside the bootstrap folder are in a "glass cage", a snapshot of the data and software.

For this reason, we could for example have a minimal bootstrap.R file that contains a single line taf.bootstrap(). As another alternative, I've created a simple README that guides users to run taf.bootstrap() before sourceAll(). The README still doesn't convey that the bootstrap procedure should not be run every time.

When and how the bootstrap is run, this is something we'll take a closer look at in the next weeks.

[arni-magnusson]

Early boot procedure discussion, obsolete in 2023.