PlanetCatfish.com

On the subfamily page for the Auchenipterinae, appear to be included.

See http://www.planetcatfish.com/common/sub ... mily_id=77.

They are still at Bagridae.

racoll wrote:On the subfamily page for the Auchenipterinae,

Rita

appear to be included.

See http://www.planetcatfish.com/common/sub ... mily_id=77.

Good spot, but how did you get there? I am unable to find a Centromochlinae page nor am I able to find the Auchenipterinae page without your link.

From any species page using the hierarchy at the top:

Cat-eLog > Auchenipteridae > Centromochlinae > Centromochlus Add this species to your My Cats species list + Add to My Cats

So, subfamily assignment is all automatic. I'm not sure what has caused this, but will try to figure out.

Edit: I reset the data for the subfamily of the genus "Rita" in my local database and reran the script that updates the subfamily, I can reproduce the problem. I think it's caused by Catalog of Fishes returning a long list of species when askign for "Rita rita", because it lists things that match "Rita" in general. It will need to be fixed by being more restrictive with regards to what is considered a "match" in the script. I will look at this tomorrow.

--
Mats

Fix for the underlying problem in this case in R1431 ("don't keep the subfamily when wrong species is found"). I'm not sure if there are any other genera that are wrong, and I suspect it's not that easy to find out, but will have a look at updating the update-script to fix that. In the meantime, if Jools sets the subfamily for Rita to 0, and reruns the script (or waits until Sunday when it runs anyway), it should be fixed.

--
Mats

Thanks Mats, the updated code has been uploaded and the subfamily_id on the genera table for the genus Rita is set to 0.

This means that Rita no longer shows up in the wrong subfamily. After Sunday, it should appear in the right one...

Jools

Have now sent a modified script to the repository for Jools to review and upload at some point. In R1432. This will remove any "subfamily" link that isn't correct. Once this script has been run (happens every Sunday at present, but it can be run manually by Jools), the four genera with problems will be corrected.

--
Mats

All fixes are now online and I'm content to wait 48 hours or so until the script now run. Basically, I think this is now resolved.

As a nice extra feature that Mats put in, subfamily links now work. Check out http://www.planetcatfish.com/mochokinae!

Jools

[Wee footnote to Mats, where the subfamily exists as a genus, this doesn't quite work as expected. e.g. Loricariinae]

Jools wrote:[Wee footnote to Mats, where the subfamily exists as a genus, this doesn't quite work as expected. e.g. Loricariinae]

I did notice that. Not sure what we should do there - they are not real, legitimate genuses, but at the same time, we use them as if they are, and "genus" has higher priority (in the "what could this name be" resolving code), so it goes for that instead. We could turn it around, but then if you click on "Loricariinae" (the genus-like symbol), it won't do the right thing [or at least not if it's a clog link like "The genus " ])

No matter what we do, aside from "rename things that are called the same as a subfamily" will solve it in a perfect way.

--
Mats

I had a look at the problem and I think the code/approach is good and the problem is a data issue. In this day and age, I think we can "do better" than just plonk something undescribed in a subfamily.

In the case of loricariinae, there are no longer any species belonging to that genus record. So I have fixed the problem by renaming and reusing the genus (That ID is now changed to be ).

Jools