they never thought I would query-replace an uppercase

[Dan Jacobson]

In todays episode a hasty Dan is again mislead by the documentation,
whereupon he stumbles upon something that needs to be documented.  All
is well in the end though.

The help for query-replace-regexp just says: "Preserves case in each
replacement if `case-replace' and `case-fold-search' are non-nil and
REGEXP has no uppercase letters."

Therefore, the user in a rush to replace a certain capital letter
thinks: Oh great, I was hoping to match only "S" and not "s", but this
possibility was apparently overlooked [by guess who].  Same apparently
with plain query-replace.

(dired-do-query-replace-regexp "S" "000") is what I was poking around
with.  Wanna know something really funny?  That command pooped out
with "replaced 0 occurrences" resting on the first match [a little
"s"].  I had to use little "s" to get the usual behavior.
echo s>file1
echo S>file2
LC_ALL=C emacs --no-site-file -q .
m m [mark all files]
(dired-do-query-replace-regexp "S" "000")
see what I mean.

By the way, I did "man ascii" to get the octal code for S, but it
seems I just couldn't figure out how to fool query-replace-regexp with
it.

Back to the first problem: from the Info page we see that we should
try setting case-fold-search to nil.  [We don't see such details
elsewhere, *Help* just sends us in my initial direction.] Didn't help:
perhaps because this becomes a local "t" variable in each new buffer
dired visits.

So you might mention on the dired-do-query-replace-regexp Help blurb
to do (set-default 'case-fold-search nil), which apparently has got me
rolling again.
-- 
http://jidanni.org/ Taiwan(04)25854780