#255 Can't "q" after "Fu"

[minshall]

if i "emacs -q" (no init file), then "Fr" (to give me a message to play
with), then "o" (to move that message to a folder), then (THIS IS KEY)
"Fu" and reply "yes" (to undo all pending changes in the current
folder),

then, if i say "q", i get "mh-quit: Wrong type argument: consp, nil"
(after colon in minibuffer; full message in *Messages*).

[minshall]

here's (what i assume to be) a backtrace:
----
Debugger entered--Lisp error: (wrong-type-argument consp nil)
mh-update-sequences()
mh-quit()
call-interactively(mh-quit)
mh-show-quit()
call-interactively(mh-show-quit nil nil)
----

[minshall]

more random:

1. the bug seems to apply only to the folder in which "Fu" was done.
2. the bug seems to go away if you "Fr" the folder in which "Fu" was done.

[Bill Wohler]

No need to lower the priority on this. If someone else hits this bug, we should definitely bump up the priority.

[minshall]

well, i ran into this again today (and, searching the bugs, found this forgotten bit!:). not that i'm lobbying to have the priority increased, especially as it doesn't seem to have happened to anyone else.

if i do "F u" (mh-undo-folder) in +outbox, the "q", i get an error

(wrong-type-argument consp nil)

in (mh-update-sequences). i think this is because

(mh-find-seq 'cur)
returns nil, which is then used (in variable seq-entry) as the target
of a setcdr, which, i assume, generates the error.

to reproduce: take a random folder, say "F u", then ":e (mh-find-seq 'cur)", and i wager the answer will be nil.

also, (mh-show-msg) invokes (mh-update-sequences), so fails while the folder is in this state. ("F r" resets 'cur).