Re: [Yaml-core] 8-bit vs Unicode (Binary Values) Revisited

[Clark C . E. <cc...@cl...>]

On Fri, Jun 15, 2001 at 02:41:31PM -0400, Clark C . Evans wrote:
| On Fri, Jun 15, 2001 at 07:39:15AM +0200, Oren Ben-Kiki wrote:
| | - Binary taken to be text. This would be *very* rare. Also,
| |   the application code dealing with binary data wouldn't 
| |   treat the value is text - it will call things like 'unpack' 
| |   etc. on it. Since the byte-by-byte layout of the value *has 
| |   not changed* (UTF-8 to the rescue), the code will still work. 
| |   The YAML file will be misleading, true. In short, a minor 
| |   (and rare) problem.
| 
| Let me present a proof that this *is* a problem.
| Consider two one byte binary strings.. \x41 ("A")
| and \xDC ("capital U with umlaut").

Ok.  Assuming that .39 of all 8 bit combinations
characters are ASCII... this means that .39^n 
documents will have this problem.  And given the
switch in type, the problem will probably not
be recoverable.  So.  For a small, 200 byte 
binary payload, the chances are... 1.1e-84

Perhaps this is significantly rare.

Clark