#DrupalSouth 2024

I understand that this issue here was raised to explicitly allow underscores in attribute names, however, I would like to point out the following from https://www.w3.org/TR/REC-xml/#NT-Name

• Attribute names should follow the NameStartChar (NameChar)* format
• NameChar may contain dashes, dots, mid-dots, numbers (among other characters), but it does NOT explicitly state allowing underscores
• also, NameStartChar (the first character of attribute names) must NOT be a dash - it can however be a colon ( : ) or an underscore.

So unless I'm interpreting that wrong, the current regex is incorrect, as it allows the first character of attributes to be a dash. So if we want to allow underscores, they should only be allowed as the first character of the attribute name.

So unless we don't want to follow the standard strictly (in order to allow the use case for which this issue here was opened), the regex should be like so instead:
[_a-zA-Z][-a-zA-Z0-9]*

The above:

• allows the first letter of the attribute name to be an underscore, but NOT a dash
• allows any character following the first character to be a dash, but NOT an underscore

Not sure about how people feel about the rest of the characters (colons, dots, mid-dots etc.). They seem edge cases to me, and best left to be discussed in a follow-up issue, in order to avoid derailing this one here.

my 2c.

I had the same issues, applied the patch from #175 and it worked correctly. It seems it is still needed even after the previous merge at #170