Re: Performance degradation from long lines

[Stefan Monnier]

> In summary, it seems that it might be possible to significantly
> improve performance by altering Emacs' display engine to use simpler
> techniques at cost of increased maintenance for the extra code that
> will be needed, and the loss of some display capability.
> I encourage reading the original thread for details.

I think it would be worthwhile to have a specific minor mode we can
enable for those cases.

> In the case of gui Emacs, the reduction of display capability is real,
> but it seems prefereable to freezing Emacs or slowing it down to a crawl.

AFAIK there are *several* unrelated features that can cause very serious
slow downs in the presence of long lines.  So we need to attack them one
by one.  A dedicated minor mode would be a good place to accumulate
the workarounds.

> I had three other ideas, which I did not yet have a chance to mention
> in the thread.

These seem to concentrate on trying to detect the slow case and partly
reduce its impact.  So, they're largely orthogonal.  I don't expect that
make-thread would make a noticeable difference, but auto-detction of
performance problems might be helpful.


        Stefan