address completion failing - how to debug?

[Ken Mankoff]

As of a week ago or so, mu address completion is failing. Not at first, but after I use emacs for a while. On a clean emacs install, tab-completion works. But now if I try to add an address, it doesn't work. Emacs flashes (visual bell?) and the minibuffer says "No match".

Does anyone have any advice how I would find out what is corrupting address completion?

Thanks,

-k.

[Dirk-Jan C. Binnema]

Please always include versions (emacs, mu) when you have questions like
these, that makes it easier to track things down.

Also, a little bit of systematic experimentation would help us, i.e.,
- how long is 'a while'
- does the same address work just after you start emacs, but not
later?
- and then work again when restarting emacs?
- did you try with just emacs -q and then loading mu4e?

Kind regards,
Dirk.

--
Dirk-Jan C. Binnema Helsinki, Finland
e:dj...@djcbsoftware.nl w:www.djcbsoftware.nl
pgp: D09C E664 897D 7D39 5047 A178 E96A C7A1 017D DA3C

[cho...@chopps.org]

So I occasionally have issues with contact tab completion as well. I
have not had a chance to debug this. I do seem to get it to work by
quiting mu4e, then starting it again. I also make sure I wait until I
see the "loaded 9432 contacts" (whatever the number is) in the
mini-buffer. Then if I try and compose an email the tab completion
works. This waiting comes from me assuming I somehow must have aborted
the contact loading when I first launched mu4e... I actually really
doubt this is correct and it's probably just the restarting of mu4e
that's fixing things. :)

Thanks,
Chris.

[cho...@chopps.org]

I'm getting more motivated to get this fixed, and started looking around
in the code. The problem is that at some point after starting mu4e when
I try and compose a new message or reply, contact completion returns no
matches. Quiting and restarting mu4e fixes the issue.

I haven't figured out what is going on; however, I do see that when I
can't complete contacts the following is true:

- mu4e~contacts is nil

- (mu4e~proc-ping) generates and error:

"error in process filter: mu4e~default-handler: Not handled: ((:crypto t :guile t :version "0.9.16" :doccount 194716))
error in process filter: Not handled: ((:crypto t :guile t :version "0.9.16" :doccount 194716))"

This might be normal but when I do this when things are working it
flips me from scratch to the main mu4e view.

Can you suggest other things to check for when I encounter the issue
again?

mu4e version: 0.9.16
emacs version: 24.5.1
archlinux

Thanks,
Chris.

[Dirk-Jan C. Binnema]

On Thursday May 19 2016, cho...@chopps.org wrote:

> I'm getting more motivated to get this fixed, and started looking around
> in the code. The problem is that at some point after starting mu4e when
> I try and compose a new message or reply, contact completion returns no
> matches. Quiting and restarting mu4e fixes the issue.
>
> I haven't figured out what is going on; however, I do see that when I
> can't complete contacts the following is true:
>
> - mu4e~contacts is nil
>
> - (mu4e~proc-ping) generates and error:
>
> "error in process filter: mu4e~default-handler: Not handled: ((:crypto t :guile t :version "0.9.16" :doccount 194716))
> error in process filter: Not handled: ((:crypto t :guile t :version "0.9.16" :doccount 194716))"
>
> This might be normal but when I do this when things are working it
> flips me from scratch to the main mu4e view.
>
> Can you suggest other things to check for when I encounter the issue
> again?

That narrows things down... what seems to happen is the that the handler
function for "pong" responses is not what it should be. This handler is
set in mu4e~start, and this apparently works when (re)starting mu4e. The
function is never un-set, but apparently (from the data above), it seems
it _is_ unset somehow. Hmmm...

Are any other errors reported in the *Messages* buffer before the time
before the symptoms appear? Are perhaps any third-party modules that
interact with mu4e involved?

[cho...@chopps.org]

So I still hit this all the time (I don't believe I have anything 3rd
party that is interfering with mu4e).

While I was playing around with some old variables in my scratch buffer
(I have it setup to not clear between restarts), I was able to have
mu4e~contacts go from nil to filled after executing the following
function:

(mu4e~request-contacts)

I'm pretty sure given that it was sitting there in my scratch that I'd
tried this before a while ago, and it didn't work, so perhaps something
changed. I'll keep checking this out and update the list with more info
as I have it.

Thanks,
Chris.

Dirk-Jan C. Binnema <dj...@djcbsoftware.nl> writes:

[Dirk-Jan C. Binnema]

On Friday Oct 28 2016, cho...@chopps.org wrote:

> So I still hit this all the time (I don't believe I have anything 3rd
> party that is interfering with mu4e).
>
> While I was playing around with some old variables in my scratch buffer
> (I have it setup to not clear between restarts), I was able to have
> mu4e~contacts go from nil to filled after executing the following
> function:
>
> (mu4e~request-contacts)
>
> I'm pretty sure given that it was sitting there in my scratch that I'd
> tried this before a while ago, and it didn't work, so perhaps something
> changed. I'll keep checking this out and update the list with more info
> as I have it.

If you find a way for others (esp. myself!) to reproduce it, I'm sure it
can be fixed. But without...

I.e., you could try to reproduce with `emacs -Q' and only a minimal mu4e
setup (like the one in the manual), and see if the problem is still
occurring.

[Stig Brautaset]

Dirk-Jan C. Binnema writes:

> On Friday Oct 28 2016, cho...@chopps.org wrote:
>
>> So I still hit this all the time (I don't believe I have anything 3rd
>> party that is interfering with mu4e).

[...]

> If you find a way for others (esp. myself!) to reproduce it, I'm sure it
> can be fixed. But without...
>
> I.e., you could try to reproduce with `emacs -Q' and only a minimal mu4e
> setup (like the one in the manual), and see if the problem is still
> occurring.

I've just started using mu4e (attempting to switch from Gnus) and
couldn't get mu4e's address completion to work until I dropped message-x
mode. Not sure if that is relevant to this case.

Stig