Matthew Lear <matt@bubblegen.co.uk> writes:

> On 03/12/16 01:58, David Bremner wrote:
>> Matthew Lear <matt@bubblegen.co.uk> writes:
>> 
>>>
>>> I'd happily post some debug info to help analyse this if somebody could
>>> give me a few pointers about what would be needed and the emacs-fu
>>> needed to generate this.
>>>
>> 
>> I think the best thing at this point would be to find (or create)
>> another public message where you can duplicate the problem.
>
> I've sent two test emails directly to David for analysis.
> --  Matt

So I _finally_ got around to looking at these, and I think it's roughly
the same shr bug as before but some different functions.

I could actually only duplicate the bug with emacs-reply-fail-ec
message, but that was fixed by the following patch against the emacs-25
branch. Does this patch fix both failures for you?

diff --git a/lisp/net/shr.el b/lisp/net/shr.el
index 6c35a33c9c..2bc37c64bd 100644
--- a/lisp/net/shr.el
+++ b/lisp/net/shr.el
@@ -1993,6 +1993,9 @@ shr-pixel-buffer-width
     (if (get-buffer-window)
        (car (window-text-pixel-size nil (point-min) (point-max)))
       (save-window-excursion
+        ;; Avoid errors if the selected window is a dedicated one,
+        ;; and they just want to insert a document into it.
+        (set-window-dedicated-p nil nil)
        (set-window-buffer nil (current-buffer))
        (car (window-text-pixel-size nil (point-min) (point-max)))))))
 
@@ -2036,6 +2039,9 @@ shr-render-td-1
            (shr-indentation 0))
        (shr-descend dom))
       (save-window-excursion
+        ;; Avoid errors if the selected window is a dedicated one,
+        ;; and they just want to insert a document into it.
+        (set-window-dedicated-p nil nil)
        (set-window-buffer nil (current-buffer))
        (unless fill
          (setq natural-width