Comments (11)

Petya Kangalova
Petya Kangalova

Steven Flower Yes, the DAC have now added the codelist in XML format, following a number of conversations we had with them. I have just sent some feedback on the codelist - one codelist was missing and the XML still requires a few changes to comply with the IATI XML schema. Once the changes have been made and we are happy that the file meets the IATI codelist schema, I will post again in here.

Andy Lulham
Andy Lulham

I’m very cautiously optimistic, but I haven’t checked or tested anything yet!

Image removed. petyakangalova:

Once the changes have been made and we are happy that the file meets the IATI codelist schema

Meeting the schema is good. Another really important check is: Does the data in the XML match the data in the XLS? Because it’s important to note: the DAC publishing XML codelists is not new – they’ve been doing that for a long time. But the DAC publishing XML that is consistent with the XLS would be new – and a really fantastic step forward.

I remember from Dale Potter ’s post on this topic:

Image removed.Planning for machine readable, version controlled OECD-DAC codelists

However, by Autumn 2017, the DAC are planning to implement an automated systems to generate both XLS and XML versions in an automated way, which should see an end to inconsistencies between versions and make it possible to work with DAC codelists in a fully machine-readable way.

If the DAC did manage to implement it in this way – with both versions generated automatically from the same source – then that would be a massive plus to helping ensure the reliability of the published XML.

Bill Anderson
Bill Anderson
Image removed. andylolz:

Does the data in the XML match the data in the XLS

It was explained to me that the xml is being generated directly from the main database and should therefore be authoritative.

Steven Flower
Steven Flower

Looking good - thanks Petya Kangalova and IATI Technical Team

Just linking up with a GitHub thread where there’s some discussion about the DAC Channel Codes list (ie the list of of Organisations originally published as an IATI codelist). If the XML output from DAC is synched up, then this can be the list for all relevant XM-DAC listings.

Bill Anderson
Bill Anderson
Image removed. stevieflow:

this can be the list for all relevant XM-DAC listings

There is a persistence problem. In DACland you can be a recipient one year and a donor the next.

Andy Lulham
Andy Lulham

I’ve made a start at comparing the XLS and XML. My work in progress findings are here: https://gist.github.com/andylolz/3962f25d7d8ef5b0675029cfe04fbf72

As mentioned, the additional flow type in the XML (38) seems like a bit of a worry.

Image removed. petyakangalova:

I have just sent some feedback on the codelist - one codelist was missing and the XML still requires a few changes to comply with the IATI XML schema. Once the changes have been made and we are happy that the file meets the IATI codelist schema, I will post again in here.

Could we compare notes? E.g. was the missing codelist this one?

Steven Flower
Steven Flower
Image removed. bill_anderson:

In DACland you can be a recipient one year and a donor the next.

But you’d persist if you were on the channel code list?

Andy Lulham
Andy Lulham

While I didn’t find any other big anomalies, there are a lot of small inconsistencies (capitalisation; spacing; small differences in translation) that suggest the source for the two datasets might be different.

Just to expand on the flow type issue: the XML includes the following two flow types:

  • 37 (Other Private flows at market terms - Banks)
  • 38 (Other Private flows at market terms - Non-banks)

The XLS doesn’t include 38, but does include:

  • 37 (Other Private flows at market terms)

So not only is 38 present in the XML but not the XLS, but I’d also wonder whether 37 might mean something different in the two versions?


Please log in or sign up to comment.