Re: Remove Ghost groups

[Eric Abrahamsen <eric@ericabrahamsen.net>]

Harry Putnam <reader@newsguy.com> writes:

> Eric Abrahamsen <eric@ericabrahamsen.net> writes:
>
>> Harry Putnam <reader@newsguy.com> writes:
>>
>>> Eric Abrahamsen <eric@ericabrahamsen.net> writes:
>>>
>>>> Harry Putnam <reader@newsguy.com> writes:
>>>>
>>>>> [NOTE: this message is probably a duplicate of a previous post where I
>>>>> forget to include the Image I intended to attach]
>>>>>
>>>>> I have developed a number of what might be called ghost groups.  Some
>>>>> (most) are as the result of removing groups behind gnus back. Others,
>>>>> I'm not sure of their origin.
>>>>>
>>>>> These groups now appear in gnus -> groups buffer with an asterisk
>>>>> before them as in the included image below.
>>>>>
>>>>> They do not respond to simple C-k nor do they respond to C-u G del
>>>>
>>>> Can you tell us the backend of the groups, and also what happens when
>>>> you hit C-k? Is there a message of any kind?
>>>
>>> When I hit C-k  the pointer goes into `working' mode and never
>>> returns.  When I force it with C-g all but one nntp group disappears
>>> and I have to press `l' to get the full setup back
>>>
>>> checking the Messages-buffer it only shows
>>>
>>>    *beep*
>>>     Quit
>>
>> Can you `toggle-debug-on-quit', do it again, and post the backtrace?
>
> Sorry for making you work too hard to help. I should have included
> backtraces from the start ...

And I should have asked -- also your Emacs version, as that's relevant.

> Included below:
> Backtrace from quiting with C-k on nnml and nnimap.  Also backtrace
> from quitting  on C-u G <bkspc> on nnimap
>
> -------       -------       ---=---       -------       ------- 
>
> Backtrace from nnml group with C-k . . .  followed by C-g
>
> Debugger entered--Lisp error: (quit)
>   gnus-group-goto-group("nnml: AdamSj\303\270gren")

This makes me think you're using an Emacs/Gnus from before my hash-table
changes (six months or so ago?), as the group strings should always be
fully decoded strings (ie there shouldn't be the escapes in the group
name). IIRC, `gnus-group-goto-group' can sometimes enter an infloop if
it can't find the group (which is a bug).

>   gnus-topic-change-level("nnml: AdamSj\303\270gren" 9 3 nil)
>   gnus-group-change-level((nil ("nnml: AdamSj\303\270gren" 3 nil nil "nnml:")) 9 nil)
>   gnus-group-kill-group(nil nil)
>   gnus-topic-kill-group(nil)
>   funcall-interactively(gnus-topic-kill-group nil)
>   call-interactively(gnus-topic-kill-group nil nil)
>   command-execute(gnus-topic-kill-group)
>
> -------        ---------       ---=---       ---------      -------- 
>
> Backtrace from nnimap group with C-k . . .  followed by C-g
>
> Debugger entered--Lisp error: (quit)
>   gnus-group-goto-group("nnimap+hput3:INBOX.Gas-South")

That doesn't really explain why this would infloop, though, as the group
name is not multibyte.

>   gnus-topic-change-level("nnimap+hput3:INBOX.Gas-South" 9 3 nil)
>   gnus-group-change-level((nil ("nnimap+hput3:INBOX.Gas-South" 3 ((1 .
> 2)) ((unexist)) "nnimap:hput3" ((modseq . "37742") (uidvalidity .
> "1495544836") (active 1 . 2) (permanent-flags %Answered %Flagged
> %Draft %Deleted %Seen $X-ME-Annot-2 $IsMailingList $IsNotification
> $HasAttachment $HasTD %*)))) 9 nil)
>   gnus-group-kill-group(nil nil)
>   gnus-topic-kill-group(nil)
>   funcall-interactively(gnus-topic-kill-group nil)
>   call-interactively(gnus-topic-kill-group nil nil)
>   command-execute(gnus-topic-kill-group)
>
> -------        ---------       ---=---       ---------      -------- 
>
> Backtrace from C-u G <bkspc> on  nnimap group followed by C-g
>
> Debugger entered: ("Quit")
>   nnimap-wait-for-response(93)

This appears to be a completely different problem? At least, I can't
immediately think of why it would be related to the two errors above.
Let's put this one aside until we figure out the previous issue.