After change-folder listing the attachments folder does not work

[Kai Weber]

Hello,

I have a quite common multi account setup with multiple folders for my
account. I use folder-hooks to switch various settings based on the
folder I change into.

I have some macros to go directly to one folder:

macro index ,i "<change-folder>=personal/INBOX<return>"

If I use this macro to go to a folder, when composing a mail adding
attachments does not work as expected:

1. Use the macro to go into a folder
2. Start composing a mail
3. try to add attachments by pressing "a"
4. If it says Attach file ('?' for list): press "?"
5. Error message is: No such file or directory =personal

It seems, <change-folder> does also change to the directory =personal
while switching folders by browsing through the mailbox list does not
change the directory and stays in my $HOME directory.

Do you experience the same problem?

You can have a look at my config on github
https://github.com/kwbr/config-mail/tree/master/.mutt

[Ian Zimmerman]

Post by Kai Weber
I have a quite common multi account setup with multiple folders for my
account. I use folder-hooks to switch various settings based on the
folder I change into.
macro index ,i "<change-folder>=personal/INBOX<return>"
If I use this macro to go to a folder, when composing a mail adding

I strongly suspect this is related to the following discussion:

https://github.com/neomutt/neomutt/issues/609

You could try neomutt, where my proposed patch has been integrated.

[Kevin J. McCarthy]

Post by Ian Zimmerman

Post by Kai Weber
I have a quite common multi account setup with multiple folders for my
account. I use folder-hooks to switch various settings based on the
folder I change into.
macro index ,i "<change-folder>=personal/INBOX<return>"
If I use this macro to go to a folder, when composing a mail adding

https://github.com/neomutt/neomutt/issues/609

Look like "your" patch contained my fixes, posted in
<***@zaogao.lan> on mutt-dev. Thanks for the (lack
of) attribution. Very classy.

[Ian Zimmerman]

Post by Kevin J. McCarthy

Post by Ian Zimmerman
https://github.com/neomutt/neomutt/issues/609

Look like "your" patch contained my fixes, posted in
of) attribution. Very classy.

I apologize.

Let's change it to "my proposal". That OK with you?

[Kevin J. McCarthy]

Post by Ian Zimmerman

Post by Kevin J. McCarthy

Post by Ian Zimmerman
https://github.com/neomutt/neomutt/issues/609

Look like "your" patch contained my fixes, posted in
of) attribution. Very classy.

I apologize.
Let's change it to "my proposal". That OK with you?

I don't really care what you call it on this mailing list; the
attribution was deserved in the commit.

I understand it was probably easy to forget, while in the midst of
dissing the "ultra-conservative" mutt devs and community, that I fixed
your patch and at least gave it serious consideration.

[Ian Zimmerman]

Post by Kevin J. McCarthy
I understand it was probably easy to forget, while in the midst of
dissing the "ultra-conservative" mutt devs and community, that I fixed
your patch and at least gave it serious consideration.

The real reason it was easy to forget was the time that had passed. But
you're right, I should have attributed you despite that. I'm flawed
like everyone.

[Kevin J. McCarthy]

Post by Ian Zimmerman

Post by Kevin J. McCarthy
I understand it was probably easy to forget, while in the midst of
dissing the "ultra-conservative" mutt devs and community, that I fixed
your patch and at least gave it serious consideration.

The real reason it was easy to forget was the time that had passed. But
you're right, I should have attributed you despite that. I'm flawed
like everyone.

Okay, fair enough. I admit I'm being a bit of an ass. For that I also
apologize.

[Kai Weber]

Post by Ian Zimmerman
https://github.com/neomutt/neomutt/issues/609
You could try neomutt, where my proposed patch has been integrated.

I just build the Github master and it works.

[Kevin J. McCarthy]

Post by Kai Weber

Post by Ian Zimmerman
https://github.com/neomutt/neomutt/issues/609
You could try neomutt, where my proposed patch has been integrated.

I just build the Github master and it works.

In an off-list reply, Kai mentioned he's running the Debian neomutt
package.

Kai, would you be willing to try compiling the mutt-1.8.3 tarball or
mercurial tip and see if the bug is present there?

[Kai Weber]

Post by Kevin J. McCarthy
In an off-list reply, Kai mentioned he's running the Debian neomutt
package.

Which was a mistake...

Post by Kevin J. McCarthy
Kai, would you be willing to try compiling the mutt-1.8.3 tarball or
mercurial tip and see if the bug is present there?

Mutt 1.8.3 tarball works.
NeoMutt Github master works
Debian NeoMutt/Mutt package NeoMutt 20170609 (1.8.3) does not work

So, I have to file a bug against the Debian package?

[Ian Zimmerman]

Post by Kai Weber
Mutt 1.8.3 tarball works.
NeoMutt Github master works
Debian NeoMutt/Mutt package NeoMutt 20170609 (1.8.3) does not work

This looks like some fix from mutt (the real thing) that has recently
been merged into neomutt, but only after debian froze their package.

[Kevin J. McCarthy]

Post by Kai Weber

Post by Kevin J. McCarthy
In an off-list reply, Kai mentioned he's running the Debian neomutt
package.

Which was a mistake...

Yes, sorry I meant that as an explanation of why I was asking you to try
the mutt-1.8.3 tarball, not as an accusation.

Post by Kai Weber

Post by Kevin J. McCarthy
Kai, would you be willing to try compiling the mutt-1.8.3 tarball or
mercurial tip and see if the bug is present there?

Mutt 1.8.3 tarball works.
NeoMutt Github master works
Debian NeoMutt/Mutt package NeoMutt 20170609 (1.8.3) does not work
So, I have to file a bug against the Debian package?

It sounds like a good idea.

However, there is currently a freeze in the Debian mutt package, until
Debian bug #870635 is resolved. You may want to wait until their next
upload and see what happens.

[Kai Weber]

Post by Kevin J. McCarthy

Post by Kai Weber
Mutt 1.8.3 tarball works.
NeoMutt Github master works
Debian NeoMutt/Mutt package NeoMutt 20170609 (1.8.3) does not work
So, I have to file a bug against the Debian package?

It sounds like a good idea.
However, there is currently a freeze in the Debian mutt package, until
Debian bug #870635 is resolved. You may want to wait until their next
upload and see what happens.

I see. I wait until the the bug is resolved and use Mutt from the
tarball.

BTW, with either NeoMutt or Mutt 1.8.3 I get

Error in /home/kai/.config/mutt/muttrc, line 13: header_cache:
unknown variable
source: errors in /home/kai/.config/mutt/muttrc

Seems I have to report another bug...

Thanks for the help

[Kevin J. McCarthy]

Post by Kai Weber
BTW, with either NeoMutt or Mutt 1.8.3 I get
Error in /home/kai/.config/mutt/muttrc, line 13: header_cache: unknown
variable
source: errors in /home/kai/.config/mutt/muttrc

By default, Mutt has a minimal configuration. I would recommend first
running:
apt-get build-dep mutt

How you configure is a personal opinion, but here are some options you
may want to pass to configure:

./configure \
--prefix=/usr/local \
--with-mailpath=/var/mail \
--enable-debug \
--enable-fcntl \
--enable-hcache \
--enable-gpgme \
--enable-imap \
--enable-smtp \
--enable-pop \
--enable-sidebar \
--enable-compressed \
--with-curses \
--with-gnutls \
--with-gss \
--with-idn \
--with-mixmaster \
--with-sasl \
--without-gdbm \
--without-bdb \
--without-qdbm

I like the newer kyotocabinet header cache, so I also would install the
libkyotocabinet-dev package and add to configure:
--without-tokyocabinet \
--with-kyotocabinet

[Ian Zimmerman]

Post by Kai Weber
BTW, with either NeoMutt or Mutt 1.8.3 I get
unknown variable
source: errors in /home/kai/.config/mutt/muttrc

Is this also the case if you build neomutt from source?

This looks like a consequence of the choices made at build time.

[Kai Weber]

Post by Ian Zimmerman

Post by Kai Weber
BTW, with either NeoMutt or Mutt 1.8.3 I get
unknown variable
source: errors in /home/kai/.config/mutt/muttrc

Is this also the case if you build neomutt from source?

Yes, Neomutt also has this problem.

Post by Ian Zimmerman
This looks like a consequence of the choices made at build time.

A plain "./configure --prefix $HOME/.local" results in the header_cache
error. Your suggestion to configure --with-kyotocabinet helped.

Thanks

[Ian Zimmerman]

Post by Kai Weber

Post by Ian Zimmerman

Post by Kai Weber
unknown variable
source: errors in /home/kai/.config/mutt/muttrc

Is this also the case if you build neomutt from source?

Yes, Neomutt also has this problem.

Post by Ian Zimmerman
This looks like a consequence of the choices made at build time.

A plain "./configure --prefix $HOME/.local" results in the
header_cache error. Your suggestion to configure --with-kyotocabinet
helped.

It was Kevin who suggested that.

Here is the command I used to build my gentoo package [1]:

./configure --disable-lua --disable-notmuch --enable-gpgme --with-gdbm
--without-mixmaster --without-tokyocabinet --without-kyotocabinet
--without-qdbm --without-bdb --without-lmdb --without-idn

I believe you simply need to configure _at least one_ of the database
backends, it is not done automatically.

[1]
rsync://very.loosely.org/foolinux/mail-client/neomutt/neomutt-20170707-r1.ebuild

[Kevin J. McCarthy]

Post by Kai Weber
I have a quite common multi account setup with multiple folders for my
account. I use folder-hooks to switch various settings based on the folder I
change into.
macro index ,i "<change-folder>=personal/INBOX<return>"
If I use this macro to go to a folder, when composing a mail adding
1. Use the macro to go into a folder
2. Start composing a mail
3. try to add attachments by pressing "a"
4. If it says Attach file ('?' for list): press "?"
5. Error message is: No such file or directory =personal
It seems, <change-folder> does also change to the directory =personal while
switching folders by browsing through the mailbox list does not change the
directory and stays in my $HOME directory.

Browsing folders using '?' does change some state: it records the most
recent directory so the next browsing starts in the same place. Just
entering the folder name and hitting enter doesn't do that.

I'll have to dig in a bit to see the problem, but some repeatable
instructions would help. Does this happen if you start mutt, hit ",i",
and then compose a message.

[Kai Weber]

Post by Kevin J. McCarthy
Browsing folders using '?' does change some state: it records the most
recent directory so the next browsing starts in the same place. Just
entering the folder name and hitting enter doesn't do that.
I'll have to dig in a bit to see the problem, but some repeatable
instructions would help. Does this happen if you start mutt, hit ",i",
and then compose a message.

It has the same effect.

=personal: No such file or directory (errno = 2)

I grepped a little bit through the manual. Maybe I have to use <change-dir> or
a combination of <change-folder> and <change-dir>?

And, as I just found out, Debian ships NeoMutt. I hope this is not a problem.

Kai