Neil Jennings writes:
|
| The abc 2.0 draft has the following
|
| %%abc-charset iso-8859-1  (or other iso code)

Well, yes, but that doesn't seem to have a well-defined scope.   Does
it apply to the whole file? If I have a text that's mixed Russian and
Yiddish (not a hypothetic case), how do I indicate which parts are in
which character set?

Also, there's a potentially very serious gotcha  with  this  sort  of
charset  indicator:  What  if  I  copy  the  file  to another machine
(perhaps via a browser, or maybe with a file-copy  program),  and  it
decides  to  rewrite the file to a native charset on the new machine.
It will, of course, translate the above %% line to the  new  charset,
but  it  will still claim that the text is iso-8859-1, and that's now
wrong.

It's sorta like the old logic-class  question  of  how  to  correctly
translate  a  sentence  like "This sentence is in English" into, say,
French.  If you translate it as "Cette phrase est  en  anglais",  you
have  the interesting problem that the original sentence was true but
the translation is false.  One could argue that  a  translation  that
doesn't preserve truth value is not really a correct translation.  Or
is it?  This is an inherent problem with self-referential statements,
as is the above %% line.

This apparently applies to the problems I've seen with  OSX.   I  use
rsync to keep a number of directories, including my web site, in sync
on a number of different machines.  This works quite  well  with  the
linux  and  *BSD  machines,  but screws up rather badly on OSX due to
some of the changes in the file system.  A later use  of  rsync  then
propagates  the  screwups back to the linux and *BSD machines.  I can
imagine a system that converts all incoming "text" files to UTF-8  or
UTF-16.   You  couldn't  very  well expect a file-transfer program to
understand all such comments in all kinds of files and translate them
properly.  It's unlikely that the programmers will have ever heard of
abc.  So on such a system, the above  %%  line  would  end  up  being
incorrect and misleading.

To subscribe/unsubscribe, point your browser to: http://www.tullochgorm.com/lists.html

Reply via email to