Re: [Openexr-devel] UNICODE support in openexr file I/O

[Bob Friesenhahn]

On Thu, 19 Jan 2006, Drew Hess wrote:
> That's a good point, but on the other hand, that just sounds like a
> bug in the application.  It's similar to the recently-discussed

It is not a bug in the application given today's definition.  By 
changing the definition, all existing applications when use OpenEXR 
will technically be broken even if they don't immediately appear to 
be.

> We should at least state in the documentation that storing UTF-8
> strings in attributes is possible and that the application should be
> prepared for this if you need to deal with files that were created by
> third parties.

If a string is in UTF-8 then there should be an attribute associated 
with that string which says so.  Otherwise there is only the 
appearance of working.  That appearance may vaporize when the 
application is used in Europe or Asia and the OpenEXR file is shipped 
to the USA.  So perhaps someone's name is stored in an attribute and 
their name uses special accented characters or is perhaps all extended 
characters.  Then when the traditional application in the USA treats 
the attribute value as normal US-ASCII (or some Windows character set) 
the result is wrong.  But if the UTF-8 string was properly translated 
to a local character set, the result may be ok because the local 
character set provided the necessary symbols.

Bob
======================================
Bob Friesenhahn
address@hidden, http://www.simplesystems.org/users/bfriesen/
GraphicsMagick Maintainer,    http://www.GraphicsMagick.org/