I just looked at a build in auspice and saw my zambia-specific build had entries assigned to New Zealand. On closer inspection it seems that one of the zambia divisions (Southern) is assigned a lat/long in New Zealand. While I can obviously delete the NZ entry and replace / update it with a Zambian one, this would break any global builds with a NZ Southern entry. So is it possible to have multiple identical named entries in the lat/long list / is there any way to filter by region / country to select the correct division? Appreciate any advice on how to resolve this. Thanks in advance!
Hi @dbridges,
unfortunately no. This is a short-coming of our format to specify lat-longs that we baked into the system long ago now knowing what was comingā¦ Our current way of handling this is suffixing ambiguous entries with shorthands for country or state like:
Palma ES
Palma BR
we need to tackle this more comprehensively, but for the time being this hopefully helps.
richard
Thanks @rneher - good to know I am not missing something obvious. Am sure you, and the rest of the team, have a 1001 things on the to-do-list so totally understand. Pretty sure Iām not going to be able to provide any real support, but if there is something I can do let me knowā¦
1 Like
Do you mean that lat/lon is assigned by a country code? And there are conflicting country codes, like for Zambia and New Zealand?
@dbridges can you post the example here?
Hi @abitrolly.
The issue is that the lat-long details are derived from the github /default/lat_longs.tsvfile. Sample metadata (country / division etc) are matched via the name. So for example, Zambia and NZ both have an area called āSouthernā. In the lat long sheet the NZ entry is called āSouthernā and the Zambian one is called āSouthern Zambiaā. Considering our samples are from Zambia I could edit the tsv file, but this would assign NZ/Southern to a Zambian location unless I remember to switch it back / edit the NZ entries.
Essentially the matching needs to be changed to take into account higher level metadata e.g. match on multiple columns i.e. Region > Country > Division > Location so that divisions / locations that are not unique between countries / regions are correctly identified. By the way I have no idea how the information is taken into the workflow so this may be a poor suggestion!
This has been on our to-do list, as you correctly identify the problems that are occurring, but we just havenāt implemented it. In the meantime, if you are focusing on Zambia there shouldnāt be any major issue if you do lose the lat/longs for the south of New Zealand!
Found the line.
It looks like an easy fix on the data level. Adding country name in a square brackets after the district name to disambiguate the records.
@james can this break anything?
It looks like itās already there on line 13747! Therefore changing your metadata so that the ādivisionā field is now āSouthern Zambiaā should do the trick.
2 Likes