Confused about confusables

[kengruven]

I'm reading about confusable characters in CE and it warns "Note: This list is not guaranteed complete! Use it as a guide only. The Unicode character set will change at times in the future, so it's on you to keep up."

I think this is a noble goal, but I'm just not sure how it's going to work, especially as specified. Where is the list? I wasn't aware of a Unicode property called "confusable". Is there an ICU function to return the current set? (Sort of: they call it "spoof detection".)

So I googled and the first hit was this page, which says that (for example) letter O and digit 0 are confusable. Does that mean I can't use "O" or "0" in a CTE key without escaping it?

I dug around in the CE source code for a while (I'm not a Go programmer) and found this. That seems to match what's in the CE documentation, but it's very different from UTS#39.

I agree in principle that allowing "spoofing"/"confusables" can be problematic for human-readable text formats, but the rules I see here are so vague I've been researching this for an hour and I still can't tell if "O" is a valid string key or not.

[kstenerud]

The issue I'm trying to solve is problems where otherwise valid symbols make it difficult for humans to see what's going on:

c1
{
    something = ［abc］
}

Is something pointing to a list or a string? Because ［ != [, it's a string. And since it doesn't contain any reserved symbol characters, it can be printed without quotes, leading to the above situation where a computer understands what it is, but a human doesn't.

This leads to the unfortunate problem of locking down a continually changing spec (Unicode) and deciding which characters are perceptually too close to characters that alter the structure of the document... I suppose maybe I'll have to go on an exhaustive search of all possible problematic characters, which will complicate all encoders, but the alternative is to allow confusing documents.