AIT reference package (ait-reference)

About

This repository contains miscellaneous data sets which are used to assemble the ait-reference package. It contains general purpose word lists, control lists, glossaries and abbreviations. It is a new package intended to consolidate into one dedicatєd location all such reference data from previous incarnations of the AIT framework — in particular, the Dragonfly project, the Deutsch Bank PBSC factory, and the IBM Tahtum factory. Reference data was scattered in multiple packages and often redundantly hard coded in compiled software; for instance the command line qotd (quote of the day), where all quotes are hardcoded in the C binary.

This is work in progress where standardisation efforts still need thought and organisation — See the strategy article where related conclusions are written up. Reuse, maintainability and availability are the KPIs of this strategy.

In addition to consolidating existing reference data, we wish to extend this repository with new datasets providing world facts and figures in various domains — see the roadmap article for the current wishlist. For instance, we have been maintaining over more than a decade, a copy of the CIA's World Factbook; this has primarily been used to create go to market country books. In the future, we would like to transform the exising factbooks into a set of YAML files managed within the scope of this repository.

This is not, and shall not be, a big data junk yard. Information collected here is analysed, sorted and structured by humans. That said, maintaining large sets of structured information requires time and effort. So all helping hands are more than welcome. In return, if YAML is not the lingua franca for such data, we are happy to maintain builds in alternate formats, be they based on markup languages (e.g. JSON, XML, …) or custom binary formats.

Build & installation

Our build is based on GNU Autotools. We practice out of source builds in a subdirectory named, by convention, out; this essentially means that all actions, beyond the first step below, are performed in the aforementioned subdirectory.

1. Initial checkout from GitHub and re-generation of GNU Autotools base scripts:

   git clone https://github.com/ISLEcode/ait-reference
   cd ait-reference
   ./aitbuild prepare
   

2. Since this is a data-only package we can directly perform an installation

   ./aitbuild install
   

If you are a package maintainer the following tasks may apply.

• Before commiting back the master branch, lint the distribution as follows:

  ./aitbuild check
  

  If all goes well you should have a distribution tarball in the out subdirectory. You coulud also create tarballs using specific archive formats and compression schemes — e.g. make dist-gzip or make dist-bzip2; this however is irrelevant in our case since we use GitHub's release mechanisme which automatically creates source code tarballs.

• Building a binary distribution has not yet been automated, this has to be done manually — we assume here we are building a binary tarball to be added to a new GitHub release and that the released version is 1.0.0. All shell commands and scripts are assumed to run under the Kornshell.

  1. Create the tarball distribution tarball — a.k.a. binary tarball, even though this package only contains data.

     ./aitbuild distrib
     

  2. Either from the command line or from a GitHub-capable frontend, create a new GitHub release tagged v1.0.0, collect input from the ChangeLog file, and append the previously created distribution tarball.

  3. The build is not fully out of source and some GNU build-time files make their way into the source tree, at the top level with configure and m4 temporary files and scripts, and in sub-directories with transient Makefile.in files. These are excluded from revision control through .gitignore. Nonetheless, cleaning up your source tree is good practice. This hasn't yet been automated, so here are the steps, from the repository's top-level directory.

     ./aitbuild clean