Re: flyspell + longlines: hang/wait

[David Reitter]

On 26 Oct 2005, at 12:00, Miles Bader wrote:

> 2005/10/26, David Reitter <address@hidden>:
>
> > Another reason not to implement soft line wrapping as actual
> > modifications of the buffer.
>
> It's a far from straight-forward change, and certainly nothing that
> will happen before emacs is released.  Richard generally dislikes the
> idea.  So if there are problems with longlines mode (which does an
> _excellent_ job generally), they should certainly be fixed.

Well, despite the _excellent_ job it's generally doing, there is a  
whole bunch of problems with longlines.

- wrong line-breaks with variable-width fonts
- incredibly slow with large files, even when resizing the frame  
(with the appropriate customization set)
  (a 1MB file causes Emacs to hang for more than three minutes (i  
gave up then) and more when the frame is made narrow. In practical  
situations, this happens when you have longlines mode turned on by  
default for all text modes (sensible for me) and then work with a  
long log file or other data.)
- line numbers displayed are plain wrong
- the interaction with flyspell (and whatnot) that has been discussed  
here.

These are known, not new problems. You will probably not care about  
it if you use mainly code and some text, often using TTY terminals to  
view the data which have a fixed width. If you write text as well,  
making good use of a window manager and using easy-to-read fonts,  
soft wrapping is a very useful mode of working, and precisely then  
the above bugs get incredibly annoying.

All of those bugs go back to the (incredible) design that implements  
a manipulation of the way data is displayed as an actual manipulation  
of the underlying data. Model-View-Controller was invented for a  
reason...