Re: noverlay branch

[Matt Armstrong <matt@rfc20.org>]

Eli Zaretskii <eliz@gnu.org> writes:

> Unless the checks are _very_ expensive, I'd prefer not to introduce
> yet another "enable-checking" knob, and instead condition those just
> by ENABLE_CHECKING.

Ok, withdrawn.  I haven't experienced extreme slowness myself.  The
problem is theoretical at this point.

Because the checks validate the entire overlay tree before and after
primitive operations (insert, erase, etc.) they do have the potential to
be worse than the old overlay implementation running with checks on
(exponential performance, etc.).  I left the --enable_checking=overlays
diffs here sould we want them later:
https://git.sr.ht/~matta/emacs/log/my/shelve/enable-checking-overlays


> Please also consider whether we need some checks to be invoked
> manually, e.g. when something goes wrong in a session that might be
> related to overlays.  E.g., a command that dumps the overlay tree(s)
> to stderr could be perhaps useful in such cases.

I believe there is code for a new lisp function that returns a
representation of the tree.  I have never used it, and I'm not sure if
it is enabled.