Re: byte-order marks

[Ludovic Courtès]

Mark H Weaver <address@hidden> skribis:

> I wrote:
>> Having slept on this, I think I agree that 'open-input-file' should
>> auto-consume BOMs.

Good.

> So what should (open-file FILENAME "r+") do?

What about doing the same as for just “r”?  I can’t think of any
reasonable scenario where this could be a problem in practice.

> Also, Unicode 6.2 section 2.6 table 2-4 says that BOMs are only allowed
> for the encoding schemes UTF-8, UTF-16, and UTF-32.  They are *not*
> allowed for UTF-16BE, UTF-16LE, UTF-32BE, or UTF-32LE.

What about this: in ‘open-file’, if %default-port-encoding is one of the
BE/LE variants, then skip the BOM logic; otherwise, check the presence
of a BOM and consume it?

Thanks for investigating!

Ludo’.