bug#1352: 23.0.60; isearch-success-function

[Juri Linkov]

> Thanks, Juri. Looks better, to me anyway.
>
> Some details you might want to consider -

Sorry for not responding to your last input after closing this bug.
I've carefully read you message as I saw it, but found no more
room for improvement.  The final change was based on your original
report and approved by Stefan.  I think further trying to achieve
the perfection will make this worse.

In particular, your latest suggestions assume that Isearch is a library of
functions with well-defined interfaces.  This is only partially true.
Rather it is an editor's feature with a set of subfeatures.  So it would
be more practical to use uniform function names that contain the same name
prefix for all related functions and variables. This will help better
recognizing the used subfeature in other features (like Dired and Info),
helping mentally connect any related function names to it.

In essence, what you suggest is using the prefix `isearch-visible'
instead of the current `isearch-filter' in the predicate names.
I see no preference for a property-based `visible' over an action-based
`filter'.  I think the word `visible' is more confusing since it can be
misinterpreted as "visible to the user" instead of "visible to isearch".
The existing name `isearch-range-invisible' has no such problem because
it tests whether the search hit is visible to the user.  So the name
`isearch-filter-invisible' connects two features and names together:

1. visible - "visible to the user"
2. filter  - "visible to isearch"

-- 
Juri Linkov
http://www.jurta.org/emacs/