Re: BOM and principle of least surprise

At 7:47 PM +0300 5/5/04, Jarkko Hietaniemi wrote:
>  > My hope for fewer options is for reading input. That is, I'd like the
>>  default encoding for all inputs and outputs to be UTF8, unless it has
>
>We tried this with perl 5.8.0 and the feedback was overwhelmingly
>negative...  if people do "print chr 0xff" they do expect one byte,
>not two.

Of course, but that's not the only way to have a single encoding for 
input and output. For example, "use utf8" (or some newly-named 
equivalent) could have effects on all file input and output in the 
scope. A different method would be to have all input and output be 
binary, but to have standardized operator overload systems (this is 
more cumbersome than the first suggestion).

What I don't want (and what we mostly have now) is a language where 
the programmer has to remember to ask "what encoding am I using" for 
every input or output command. If I have a text processing program, 
it is likely that all input and output will be in my chosen encoding; 
those that aren't need to be read/written using different tools (such 
as subroutines that have the new encoding specified for their scope).