please tell users right way to undo inside query-replace

please tell users right way to undo inside query-replace

[Dan Jacobson]

The "?" help for query-replace mentions many things,

Type Space or `y' to replace one match, Delete or `n' to skip to next,
RET or `q' to exit, Period to replace one match and exit,
Comma to replace but not move point immediately,
C-r to enter recursive edit (C-M-c to get out again),...

But it doesn't mention how to even one little undo.

One little undo [C-/] is all it takes to break a, say,
dired-do-query-replace-regexp or other long and involved job.  This
will cause one to need to restart all the way back on the first file,
or the top of a single file.

Therefore that Help information should mention that to do a undo, one
should enter a recursive edit to do it safely.
-- 
http://www.geocities.com/jidanni/ Taiwan(04)25854780

_______________________________________________
Bug-gnu-emacs mailing list
Bug-gnu-emacs@gnu.org
http://mail.gnu.org/mailman/listinfo/bug-gnu-emacs

Re: please tell users right way to undo inside query-replace

[Richard Stallman]

    But it doesn't mention how to even one little undo.

    One little undo [C-/] is all it takes to break a, say,
    dired-do-query-replace-regexp or other long and involved job.

The doc string says,

    If you exit (\\[keyboard-quit], RET or q), you can resume the query replace
    with the command \\[tags-loop-continue]."

Did you try that?  Did it work?

_______________________________________________
Bug-gnu-emacs mailing list
Bug-gnu-emacs@gnu.org
http://mail.gnu.org/mailman/listinfo/bug-gnu-emacs

must reposition cursor for tags-loop-continue on query-replace E C-g

[Dan Jacobson]

RMS> The doc string says,

RMS>     If you exit (\\[keyboard-quit], RET or q), you can resume the
RMS>     query replace with the command \\[tags-loop-continue]."

By the way, use the E command to edit a replacement string.  Now
change your mind about editing the replacement string by hitting a
C-g.  Notice that the query-replace is now over too.  Why not have a
C-g during an E return us to the state before we hit E, whereupon if
we really wanted to bust out of the query-replace, we could hit
another C-g.

OK, one hits a C-g within an E and finds the query-replace party is
now over.  OK, one types M-x tags-loop-continue [odd, "tags", a
user who has used query-replace but never used tags would be scared
off by that name ["isn't that part of the advanced level?  That can't
be what I'm looking for.  I'm looking for a query-replace-* command
that will get me back in the loop"].

OK, we have hit tags-loop-continue, _but_: our butt had better be in
the file where the fun ended [is this mentioned?], and not say off
looking at documentation, _or_/_and_ at least the cursor must be
repositioned before our E fiasco, or else we will not be asked about
that spot... only to regret it years later, "how could I have missed
it?".

Maybe RMS just patched this... however I will wait for the next
edition instead of patching.
-- 
http://www.geocities.com/jidanni/ Taiwan(04)25854780