Alphabetical Sorting Out of Sort (While NO special characters involved)

I have something weird. I guess it’s a bug? It’s weird is what it is.

I have a list of SKUs and trying to sort them alphabetically. Simple right?

Something involving a character and a number combination (ORG vs ORG19) in the middle of the sku makes the sort misbehave. I tried some testing, but simpler use cases seem ok.


  1. How is this possible?
  2. What can I do about it?

No comment? Am I imagining this? Is this not considered an issue?

I’m puzzled. Noone else has this demonstrable issue? Or am I crazy? Can someone confirm or deny? Does anyone even see this?

Dear @cblock,

Would you mind to share a dummy copy of the doc, it’s in general more engaging for a community members to see if we can get you supported.

Thanks for the suggestion @Jean_Pierre_Traets. I really appreciate it.

1 Like

Quickest Summary: If we have thing_1 and thing1_1, Coda sorts them in the Wrong Order.

Something about number_number messes it up.

Here’s a doc to play:

Let’s see if this helps drive the point home :thumbsup:


I didn’t feel a doc would be useful as this is not a document specific issue. I am not troubleshooting a doc… I am raising a coda-wide possible :bug: that seems to be large that I don’t even understand.

Also, the issue is only visible when you compare the behavior to other systems (Sheets, Excel…), which is why I thought screeshots comparing the behavior were the way to go.

Thank you for raising this question @cblock!

I’ve posted about this internally so that we can check it out.

Looks like it’s not Coda that’s messed up — that’s everything else that’s messed up.

Coda does sorting by ASCII/Unicode codes, with no smartypants logic. In ASCII, _ comes after 1:
image

Sheets etc do some other kind of sorting/collation. Found this test doc:

Didn’t find what the collation it used though. Now curious myself.

I just posted the opposite. That was wrong.
I’m browsing myself to find out formal conventions.

I think however, the point is actually valid.

Found something:
http://demo.icu-project.org/icu-bin/locexp?_=en_US&d_=en&x=col

Apparently Sheets etc use simply use Unicode collation algorithm (some locale of it), where _ indeed comes before numbers:
image

Yeah, guess Coda should get to that eventually when they add better localization/internationalization support.