Comments (5)

Rolf Kleef

It raises the issue of versioning non-embedded codelists. An organisation could have published activities in the past, with DAC codes that were valid then. What would IATI guidance be: update the sector codes when (re)publishing the activity? Live with codes that are not on the codelist? Adapt the standard and add a value-date attribute similar to that in currencies, to be able to find the DAC list that was in use when the sector codes where added?

A quick lookup of the first four on the list to be removed shows some 8134 activities that would basically become invalid: http://www.openaidsearch.org/?sectors=15120,15140,15161,15162&offset=50&order_by=-total_budget

Mark Brough

[~467] - indeed, this makes a lot of sense. My understanding is that previously, when codes were deleted from the DAC sector code list, they were not deleted from the IATI code list so that it would still be possible to look up against them. That becomes particularly important if publishers are discouraged from including the text corresponding to a particular code. But yes, it does raise a wider issue of versioning of codelists (probably also including embedded codelists). I could see that his would become particularly problematic if an old code was re-used with a different meaning.

I’m suggesting the deletion of codes is moved to the decimal process so that any potential negative implications could be explored / discussed further there.

Ben Webb

Yes, I would advise against removing codes from the IATI NonEmbedded lists, as they are valid for historical data. In the case of DAC, the codes suggested for removal are needed to describe historic data (ie. data about projects that took place in the past).

My proposal for dealing with this problem better is at https://github.com/IATI/IATI-Codelists-NonEmbedded/pull/51 and there’s some additional information at https://iati.titanpad.com/94

Rory Scott

Hello All,

This is just a heads up to say that we’ve taken on board the above, as well as contributions for the original consultation page, and have made a proposal on the original consultation page for this issue.

This proposition involves:

  1. Adding the currently missing codelist, as well as,
  2. Updating the names and descriptions that are currently out of sync.

These two steps are linked to two separate consultations, so please feel free to had feedback to both.

Finally, do deal with the eventual glut of codes that are no longer in use, a consultation to the upcoming decimal upgrade to 2.02 that suggests adding a ‘__withdrawn__’ flag or attribute to the codelist item.


Please log in or sign up to comment.