jr@LF-SERVER-2.BBN.COM (John Robinson) (11/12/87)
The problem is not with query-replace, but with the way spell-region calls it. Spell.el notices when any words from the spell filter differ only in capitalization, but unfortunately the capitalization of the first occurence is the one used in the query-replace, so if it is mixed case, the wrong things happen. This can be fixed two ways: 1. Modify (create your private version of) the filter spell to include a lower-casifying step: change the line like deroff -w $F | sort -u | /usr/lib/spell to deroff -w $F | sort -uf | /usr/lib/spell This seems (at least on the Sun) to preserve only the all-lowercase variant. 2. Do the following in spell.el: change the expr (setq word (buffer-substring (point) (progn (end-of-line) (point)))) to (setq word (downcase (buffer-substring (point) (progn (end-of-line) (point))))) I'd vote for the latter to be adopted in spell.el, since the original behavior of /usr/bin/spell might be critical elsewhere. Of course, there is always ispell... /jr ------- Forwarded Message Date: Tue, 10 Nov 87 09:41:00 -0400 From: Steve Azmier <STEVE%YULIBRA.BITNET@WISCVM.WISC.EDU> To: jr@BBN.COM Subject: query anomaly Thanks John...I agree that is how query-replace works...unfortunately the context I was using it when the problem occurred was the spell.el program. Since it flags capitalized mistakes first (it is alphabetical - ascii order), I am left with the problem of having the FROM-String capitalized, so in fact the query-replace is not designed to handle this case. Regards Steve ------- End of Forwarded Message