Re: highlight-changes-mode

[rsharman@pobox.com]

Richard Stallman writes:
 >     I added a new keyword :only-new EXPR which allows restricting
 >     a change of mode to new only new buffers if EXPR is non-nil.
 >     (So the default behaviour, without this keyword, is unchanged.)
 > 
 > If we want that functionality, this is a clean way of providing it.  I
 > think the real question is which way we want global minor modes to
 > work.  Do we want the global mode command to affect only buffers
 > created subsequently?  Do we want it to affect all existing buffers?
 > Only existing buffers in which the user has not specified any
 > setting for the mode?
 > 
 > Is there a good reason why global-highlight-changes should
 > not be like Global Font Lock mode, and affect all buffers
 > where it is applicable?


That wouldn't be a problem when the global mode is turned _on_;
the problem occurs when the global mode is turned _off_.

The current implementation allows the situation where a user no longer
wants it turned on by default, but has an existing buffer that has
some changes already highlighted.  If disabling the global mode
automatically disables highlight changes modes in that buffer, then
the highlighting of the changes disappears; we have lost information.

I don't remember if that _was_ the reason why it was done that way;
that's the only reason I can think of for keeping that behaviour.

One could argue against this, however, by noting that there are now
[in the new version] two commands to markup the changes; one works by
comparing against a file, the other against a buffer.  So in some
cases - probably most - this "lost" information could be recovered.