[Erc-discuss] bug in erc-stamp.el

[John J Foerch]

Hello,

I found a bug concerning the width of the erc-timestamp fields in erc-stamp.
To see the bug:

1.  Enable erc-timestamp-mode.

2.  (setq erc-insert-timestamp-function 'erc-insert-timestamp-right).

3.  Then place point in column 0 after a line that contains a timestamp.
    Press up (previous-line), and point goes to the end of the text on the
    previous line instead of staying in column 0.

I believe this is because the erc-timestamp field includes the newline
character at the end of the line.  Thus, when point is in column 0 on the
following line, and a movement command is issued, point is already effectively
inside the erc-timestamp field, and it is inhibited from moving past the left
boundary of the field.

I attached a patch to exclude the newline from the erc-timestamp field.  Note
that the newline must still be 'intangible in order to achieve the correct
effect with point movement.  I wonder though at whether this is a wise use of
fields, given how they affect point movement.  If you set
erc-timestamp-intangible to nil, the existence of a field affects the movement
of point strangely.

--John

Index: erc-stamp.el
===================================================================
RCS file: /cvsroot/emacs/emacs/lisp/erc/erc-stamp.el,v
retrieving revision 1.6
diff -r1.6 erc-stamp.el
269,270c269,270
<       (erc-put-text-property from (1+ (point)) 'field 'erc-timestamp)
<       (erc-put-text-property from (1+ (point)) 'rear-nonsticky t)
---
>       (erc-put-text-property from (point) 'field 'erc-timestamp)
>       (erc-put-text-property from (point) 'rear-nonsticky t)