Re: [AUCTeX] word-search-* and re-search-* does not work with folding

[David Kastrup]

Ralf Angeli <address@hidden> writes:

> * David Kastrup (2006-09-10) writes:
>
>> Ralf Angeli <address@hidden> writes:
>>
>>> And I always thought that `M-x re-search-forward RET' would be as
>>> interactive as `C-M-s' ...
>>
>> Sigh.  I don't mean "interactive" as in "(interactive)" here, but
>> "interactive" as in the "i" of "isearch".  So please replace
>> "non-interactive commands" in the line above by "non-interactive
>> searches".
>>
>> I would have thought it obvious that this is what I meant, but you are
>> right that I used a slightly wrong expression.
>
> Uh, the "i" in "isearch" stands for "incremental".

Cough, cough.  Indeed.

> Anyway, it's obvious now that the isearch commands invoked via
> `C-M-s RET' or `C-s RET C-w' are not equivalent to the similarly
> looking `re-search-forward' and `word-search-forward' commands.

Hm, yes.  I searched myself silly to see where preview-latex opens up
overlays when doing isearch.

It does not use search-invisible at all.  Oops.  It used to do this at
some point of time.

After ruling out the possibility that preview-latex does this by
magic now, I finally figured that it depends on the setting of

    disable-point-adjustment is a variable defined in `C source code'.
    Its value is nil

    Documentation:
    If non-nil, suppress point adjustment after executing a command.

    After a command is executed, if point is moved into a region that has
    special properties (e.g. composition, display), we adjust point to
    the boundary of the region.  But, when a command sets this variable to
    non-nil, we suppress the point adjustment.

    This variable is set to nil before reading a command, and is checked
    just after executing the command.

    [back]

Interactive^W Incremental searches set this variable, and it seems
like C-s RET does not reset it.

preview-latex will open overlays rather than skipping them if
`disable-point-adjustment' is set.

Smooth.  The XEmacs version instead hooks itself into
isearch-highlight.  I would expect that this will _not_ show a
difference between
M-x search-forward RET
and
C-s RET
but, well, it still does.  My theory would be that XEmacs still calls
isearch-highlight, but
M-x debug-on-entry RET isearch-highlight RET
throws a completely irrational error, and I ran out of energy checking
this.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum