bug#20543: 24.5; <SPC> in ispell-buffer accepts spelling for the whole line

[Marcin Borkowski]

On 2015-05-10, at 18:05, Jürgen Hartmann <juergen_hartmann_@hotmail.com> wrote:

> Using GNU Emacs 24.5.1 (x86_64-unknown-linux-gnu, GTK+ Version 3.10.2)--but
> this applies to former Emacs versions also--I found the following puzzling
> behavior of ispell-buffer:
>
> If a suspicious word is accepted once by pressing <SPC> in an interactive
> ispell-buffer session, all further occurrences of the same spelling on the
> same line are skipped, i.e. considered as accepted too.
>
> Here is an example:
>
> Open an Emacs 24.5 session (it is the same with Emacs 24.4) by
>
>    LC_ALL=C emacs -Q
>
> and enter the following line (with or without newline does not matter) in the
> *scratch* buffer:
>
>    The term charset is short for charset.
>
> Assume that the last word is a typo that should read "character set". Now
> change the dictionary to american and run ispell-buffer. The first occurrence
> of "charset" gets highlighted, but since it is correct here, we use <SPC> to
> accept it once and proceed. But oops... the spell-check finishes immediately
> without giving us the chance to correct the second occurrence of "charset" in
> that line.
>
> This finding was shortly discussed in help-gnu-emacs
> (http://lists.gnu.org/archive/html/help-gnu-emacs/2015-05/threads.html#00134)
> and it was Eli Zaretskii who indicated that there is the following comment in
> line 3651 of emacs-24.5/lisp/textmodes/ispell.el:
>
>    ;; Do not recheck accepted word on this line.
>
> This suggests that there might be a reason for that behavior. If this is
> true, what is it?

I've just seen this report.  I'm also very curious about that reason.

> Apart from that, I can not imagine why such a behavior might be desirable.
>
> Juergen


-- 
Marcin