Re: diary-redraw-calendar clobbers point in diary-file buffer.

[Glenn Morris]

Matt Hodges wrote:

> This save-excursion can't do anything meaningful to preserve point in
> the calendar buffer can it? 

No, it's complete rubbish! As you and Stefan pointed (ahem) out, I'm
saving point in the calling buffer, not the calendar. Being extra dim
this week, sorry.

But I don't understand what's going on here.

The final calendar-cursor-to-visible-date in redraw-calendar should
mean that no saving of point is necessary (once the save-excursion is
deleted completely, that only messes things up).

Does anyone have a recipe for reproducing a bug from a vanilla Emacs,
using the version without the save-excursion in redraw-calendar? I
myself cannot get any kind of error to occur (though I don't doubt
there is one).

The only problem I can reproduce is that redraw-calendar called from a
buffer other than the calendar moves the cursor to point-min in the
calendar. And yet

(with-current-buffer calendar-buffer (point))

*when evaluated in the scratch buffer* still reports a number > 1. In
other words, the cursor position does not seem to correspond to point.
But then using M-: to evaluate the above expression changes the answer
to 1. I think I'm going mad...

> The calendar buffer will be current when redraw-calendar is called
> from mark-diary-entries.

Well, probably, but not necessarily. mark-diary-entries is an
interactive function, so I can call it from any buffer I like, not
just the calendar. And as I said above, I don't see why explicitly
saving point is even necessary, given what redraw-calendar does/should
do.


I'll revert the problematic change soon if I can't figure this out.