bug#39977: 28.0.50; Unhelpful stack trace

[Madhu]

[Sorry for the delay in checking my email until now. I wasn't
subscribe to the debbgugs newsgroup]

* I didn't get Eli's message #8 (Sat, 07 Mar 2020 20:50:42 +0200) in
my mailbox.  The patch supplied in this message does indeed make the
crash go away.

* I did get Eli's message #11 (Fri, 13 Mar 2020 11:55:26 +0200) to my
mailbox. Subsequent messages seem to be delivered to my email address.

* Re Martin's message #14 (Fri, 13 Mar 2020 17:28:53 +0100), the check
(f == sf) in delet_frame is not true when I trigger the crash.

* Re Eli's message #23 (Sat, 14 Mar 2020 12:10:09 +0200): It seems out
that I am going out of the way to trigger the crash - I may be
introducing a "bug" in SLY code, or exposing a defect in SLY design.

Presumably under normal circumstances the crash should not occur.

I haven't understood the sequence of events which causes leads to this
crash case.  I'm a little embarrased to reveal it but I can try to
pass on the recipe to Martin.

* Re Martin's message #35 (Sun, 15 Mar 2020 18:49:21 +0100), the patch
frame.c.diff does make the crash go away.