Jamie,

On Mon, 20 Jun 2011 08:16:00 -0700, Jameson Graef Rollins <jrollins@finestructure.net> wrote:
Non-text part: multipart/signed
> On Mon, 20 Jun 2011 09:15:06 +0200, Sebastien Binet <seb.binet@gmail.com> wrote:
> > recently, using the head of notmuch-git:
> > 
> >  git://notmuchmail.org/git/notmuch
> > 
> > when a mail ends with an url and is recognized as a signature:
> ...
> > I can't hit [space] to go to the next thread and I get the following
> > error:
> >  End of buffer
> 
> Hey, Sebastien.  I'm guessing that it is probably the *next* message in
> the thread that is the problem, not the end of the one that you're
> seeing.
I (naively) don't think so:
I don't get the error (which looks like that in the *Messages* buffer:
notmuch-show-advance-and-archive: End of buffer
)
if I unravel the signature:
      [ 4-line signature. Click/Enter to hide. ]
      _______________________________________________
      Message sent via/by LCG Savannah
      http://savannah.cern.ch/

=> all is fine.

it is only when the signature is lumped like below, that I get the error:
      [ 4-line signature. Click/Enter to show. ]

also, it's only happening to the last message of a thread.

>  If you can determine the message id's of this message and the
> next one in the thread it will probably help with debugging.
> 
> If you haven't yet, it usually helps in debugging to view the messages
> in the terminal?  You can copy the thread-id into the clipboard by
> hitting "c i" when the cursor is over this thread in the notmuch search
> buffer, and then you can view the thread on the command line with:
> 
>   notmuch show thread:<thread-id>
> 
> It might be hard to see if there are any problems in the output, but you
> can at least see gmime is throwing an error by redirecting stdout to
> /dev/null:
> 
>   notmuch show thread:<thread-id> >/dev/null
> 
> You can try determining the message-ids with something like:
> 
>   notmuch show thread:<thread-id> | grep id
> 
> You can then view the raw messages with:
> 
>   notmuch show --format=raw id:<message-id>
> 
> If you feel comfortable doing so, it might help to forward the suspect
> message to the list.

I tried all the above, to no avail (ie: nothing on stderr.)

> 
> > this does not show up when using the 'release-candidate/0.6' branch of:
> >  git://finestructure.net/notmuch
> 
> The release-candidate/0.6 is a bit out-of-date at the moment, as most of
> it's features have been incorporate into notmuch/master, and I doubt
> that the ones that haven't would be related to the issue you're
> seeing.

hum... I could try to bisect (but I've never done that before!)

-s