- From: Ned Freed <Ned.Freed@INNOSOFT.COM>
- Date: Tue, 01 Jul 1997 10:44:44 -0700 (PDT)
- To: "Martin J. Duerst" <mduerst@ifi.unizh.ch>
- Cc: Walter Ian Kaye <walter@natural-innovations.com>, ietf-charsets@INNOSOFT.COM, IETF Languages <ietf-languages@uninett.no>
> > Here is my own personal definition (which I came up with about a week ago): > > > > A "control" character is one which can be interpreted as an instruction > > to treat the characters which follow it in a different way than they > > would be treated if they did not follow that character. > Not that useful, probably. First, many control characters don't > really treat following characters in a different way, unless you > consider the fact that they appear on a different line, for example, > as "treating in a different way". But then you would have to > attribute a "control" function to each and every character because > otherwise, they all will have to appear on top of each other :-). > Also, the property of introducing state, implied in your definition, > is something that may be in some cases unavoidable (e.g. complex > BIDI cases), but is not in general desired, just to the contrary. There are also cases where control characters modify preceeding characters in some way. However, as far as the idea of state goes, I doubt very much that you'll get agreement that state isn't desireable. I do agree that this is not a useful definition. Ned --Boundary (ID uEbHHWxWEwCKT9wM3evJ5w)
Received on Sunday, 20 July 1997 14:15:33 UTC