Corner cases (was: Re: UTF-16 Encoding Scheme and U+FFFE)

Shawn Steele Shawn.Steele at
Wed Jun 4 13:10:05 CDT 2014

The BOM I've seen (not FFFE though), it's prevalence depends on the system and other factors. 

The others I only see if there's corruption, bugs, or tests.  The most common error I see that causes those is when some developer calls a binary blob a unicode string and tries to shove it through a text transport or something.  Usually that bites them sooner or later.


-----Original Message-----
From: Unicode [mailto:unicode-bounces at] On Behalf Of Doug Ewell
Sent: Wednesday, June 4, 2014 11:01 AM
To: unicode at
Subject: Corner cases (was: Re: UTF-16 Encoding Scheme and U+FFFE)

How common is it to see any of the following in real-world Unicode text, as opposed to code charts and test suites and the like?

1. Unpaired surrogates
2. Noncharacters (besides CLDR data)
3. U+FEFF at the beginning of a stream (note: not "packet" or arbitrary cutoff point)

I'm not asking whether any of these are recommended or "prohibited" or whether they are a good idea. I'm asking about actual usage.

Doug Ewell | Thornton, CO, USA | @DougEwell

Unicode mailing list
Unicode at

More information about the Unicode mailing list