On Mon, Sep 18, 2017 at 2:51 PM, Adam Roach <adam@xxxxxxxxxxx> wrote:
On 9/18/17 16:48, Ted Lemon wrote:
Look, i don't want to get into a long exchange about this. I agree with Carsten: keeping the BOM is a past-looking choice, not a forward-looking choice, and it does cause problems.
I don't want to get in a protracted argument either, but I find your assertion that "it does cause problems" to be lacking concrete examples. Can you provide an existence proof of one such example? If you're right, it shouldn't be too hard.
/a
As a simple test case, some BCPs are displayed by concatenating the relevant RFCs. As RFC 3629, Section 6 notes, that concatenation will take U+FEFF and convert it from being a signature to being a non-breaking space. Can we test our tools to confirm that we are stripping it correctly in this case? If we're not doing it, I doubt anyone else is....
Most of the text in RFC 3629 seems still on the mark, not that I'm an expert, but if we take it to be true, it has some advice. That says that you should forbid the use of U+FEFF as a signature when you have character encoding identification mechanisms, and that you can retain it otherwise. What do we think the case is here?
regards,
Ted