> it seems that after the smerge-remove-props call, the match data
> information is lost so (match-end n) returns nil, causing
> delete-region to fail. Changing the definition as below seems to fix
> the problem.
> (defun smerge-keep-n (n)
> (let ((zero-start (match-beginning 0))
> (zero-end (match-end 0))
> (n-start (match-beginning n))
> (n-end (match-end n)))
> (smerge-remove-props zero-start zero-end)
> ;; We used to use replace-match, but that did not preserve markers
> so well.
> (delete-region n-end zero-end)
> (delete-region zero-start n-start)))
That's odd. I can't see any place in smerge-remove-props where the
match-data might be clobbered.
And I can't reproduce your problem: M-x smerge-keep-current works just fine
in my tests. I suspect it may be something like a bad interaction with an
after-change-function that doesn't properly save the match data. Can you
give a more precise recipe to reproduce the problem, starting from "emacs -Q"?
Stefan
_______________________________________________
emacs-pretest-bug mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/emacs-pretest-bug
