PlanetCatfish.com

Hi Jools,

is suffering from the same Distribution occurrence disorder that afflicted . I checked the Type locality data on the C. meridionalis CLOG page and discovered the same errors in its Lat/Long data as you found for B. minerim:
11°25’33.1'S, 55°16’39.3'W, instead of
11°25'33.1''S, 55°16'39.3''W.

...And as with B. minerim, this error places a map tag out in the middle of the Arabian Sea for the Distribution occurrences map.

I submitted a correction for this in the Type locality data field on the CLOG page, but I don't think this correction will fix the map coordinates on the Distribution occurrences map. Based on what you said before, I suspect the map coordinates are being extracted from an external database, not directly from the CLOG page info. Therefore, I don't expect that my fixing the typography of the CLOG page will then correct the map. If it does, then the problem is fixed (we should be so lucky! ) Otherwise, it will require more of your attention.

Cheers, Eric

If a data submission contains lat/long information then the stage of me vetting the info will refresh the map data. In short, your change should fix the problem. I've accepted the change now.

Jools

Yes, the map is fixed. So from what your saying, the lat/long data is in fact plotted according to whatever data appears in the Type locality data field of the CLOG page. Is that correct? In other words, the gold map flag with the star is plotted using this data, not an external source?

Okay, I understand. If I find future mistakes, I can fix them myself. Good to know.

Wait, no. That's not what you're saying. You're saying that if the original external data are flawed, it won't get caught. But if I edit the CLOG data, then when you approve the data (vet it, to use your words), then you will manually correct the lat/long data for the map. Is that more accurate?

Thanks again, Eric

There is a bit of code that runs when Jools vets the update, which tries to parse the coordinates out of the type locality field. It doesn't ALWAYS work, especially if there's something a bit weird in there. It's REALLY not easy to write code that does this. But it's better than expecting you, me or Jools to do it by hand, because MOST of the time, if you just take what it says in the original description, it comes up right. The whole function to do this is about 280 lines long, so it's not exactly a simple piece of code.

If the data is wrong when entered (including when we automatically take in data from Fishbase and CoF for things like type locality), and the data is somehow incorrect, the code that tries to understand it will either not understand it at all, or get the north/south, east/west or something else wrong.

If Jools spots that it's wrong, he can edit the field when vetting, but of course, that relies on human fallibility, so doesn't work always.

--
Mats

Mats, thanks for the explanation.
Cheers, Eric