It would be useful if one could use bookmark-set on mu4e messages, but we can't at the moment!

Hi,

I am trying to build mu 0.9.8.4-dev3 on OS X 10.7.3 (I use homebrew as package manager).

Using `./configure I get:

./configure: line 15998: syntax error near unexpected token `GLIB,'
./configure: line 15998: `PKG_CHECK_MODULES(GLIB, glib-2.0 >= 2.14 gobject-2.0)'

I do have glib-2.32.1 installed via homebrew, though.

Since homebrew won't install autoconf's libs properly (becasue of xcode), it says:

you'll need to add its lib & include paths to your build variables:

    LDFLAGS  -L/usr/local/Cellar/autoconf/2.69/lib
    CPPFLAGS -I/usr/local/Cellar/autoconf/2.69/include

Following the help section ./configure -help, i tried to add:

./configure LDFLAGS  = /usr/local/Cellar/autoconf/2.69/lib CPPFLAGS = /usr/local/Cellar/autoconf/2.69/include

but it does not work:

configure: WARNING: you should use --build, --host, --target
configure: error: invalid variable name: `'

I tried also adding:

export LDFLAGS='-L/usr/local/Cellar/autoconf/2.69/lib'
export CPPFLAGS='-I/usr/local/Cellar/autoconf/2.69/include'

and

CPPFLAGS=-I/usr/local/Cellar/autoconf/2.69/include
LDFLAGS=-L/usr/local/Cellar/autoconf/2.69/lib

I am not too experienced with configure. What is wrong with those parameters?

Thanks!

I've got an interesting (but, I think, common) use-case -- not knowing how the internals of mu work, I've got no idea whether or not this would be easy to do. Like many people, I've got a handful of (big!) mail folders that I can be certain will never be updated again -- storage/2004, storage/2005, etc. I know that if I index them, then add a .noindex file to their root, then they will stay in the index but not be scanned again in the future. But then, if I run a mu index --rebuild, I lose those messages. What I'd love to see would be a .noupdate file, similar to .noindex. If there's a .noupdate file in a Maildir, then if there are any messages from this folder already in the index, the folder won't be scanned on a mu index. During a mu index --rebuild, though, there aren't any messages from this folder already in the index, so it will get scanned once. (Something like .scanonce would also be a good name for this file.)

Would that be a plausible thing to implement, or does it run afoul of the indexing architecture somehow?

So, I recently updated mu/mu4e to find, to my joy, that you'd implemented contact auto-completion! I was previously doing this using the org-contacts method which was decent but tedious to maintain. So, I returned mu4e-org-contacts-file to nil and set off to try the new completion.

Only, it kept using org-contacts! Even when I removed my contacts file, it still tried to use org-contacts. Even after removing any (require 'org-contacts) reference, it persisted! I actually had to remove org-contacts entirely from the load path to use the new completion.

Hi,

When I run mu4e-update-mail-show-window it appears to do nothing.

I was under the impression it would update the index?

Is that correct?

I have mu4e-get-mail-command set to "true" as I don't use offlineimap

Using 0.9.9-dev4 which I pulled just now.

Thanks,

Jason

I'm reporting a configure failure on OSX 10.7.4 using XCode-4.3.2 where mu thinks xapian-1.2.10 is not recent enough. I'm updating the mu formula for Homebrew. Because we will be deploying xapian-1.2.10 in the next few days, we will be patching mu configure until a new version is released that contains a bug fix.

checking for xapian-config... xapian-config
configure: error: *** xapian version >= 1.2 needed, but version 1.2.10 found.

This is the patch we will be using that fixes the issue, albeit not so elegantly. I've tested it with success.

--- a/configure 2012-05-08 04:26:10.000000000 -0700
+++ b/configure 2012-05-11 23:20:57.000000000 -0700
@@ -17640,7 +17640,7 @@
   xapian_version=$($XAPIAN_CONFIG --version | sed -e 's/.* //')
 fi
 case $xapian_version in #(
-  1.[2-9].[0-9]) :
+  1.[2-9].[0-9]*) :
      ;; #(
   *) :
     as_fn_error $? "*** xapian version >= 1.2 needed, but version $xapian_version found." "$LINENO" 5 ;;

Because I'm not familiar with autoconf syntax, I didn't feel confident enough to fork mu and submit a pull request that patches configure.ac, but I'm sure it's easy enough. Ok thanks for considering this.

Apple Mail inlines different kinds of attachments, e.g. PDF documents. It would be great to see them indexed and flagged as attachments too.

I managed to include them with mu4e by changing parts of mu4e~view-construct-attachments-header[1], but to make it perfect they should be searchable.

[1]

(defun mu4e~view-construct-attachments-header (msg)
  [...]
  (let* ((id 0)
         (attachments
          (remove-if
           (lambda (part)
             (and (string-match "^[0-9]+\.part$" (plist-get part :name)) ;; <= exclude parts named 0.part, 1.part, ...
                  (not (member 'attachment (plist-get part :type)))
                  (not (string-match "^image" (plist-get part :mime-type)))))
           (plist-get msg :parts)))
  [...]

Wrapping lines via mu4e-toggle-wrap-lines creates a wall of text with quoted text. There are two possible workarounds I can think of:

1. Set the fill prefix to >
2. Use longlines-mode

Hi,

I was indexing new mails (the whole gnu-emacs-sources and emacs-orgmode lists, converted in maildir format, and mu returned error message mu: normalized text doesn't fit :/

$ mu index
/ processing mail; processed: 1650; updated/new: 1301, cleaned-up: 0mu: normalized text doesn't fit :/
/ processing mail; processed: 59046; updated/new: 58697, cleaned-up: 0
\ processing mail; processed: 59046; updated/new: 0, cleaned-up: 0

I could not find anywhere an explanation about it. Is that something I should worry about? When I use mu4e, it sems everything is working fine.

PS: If anybody is interested in those lists in maildir format, please let me know. I can put a compressed archive somewhere.

This would be easy to do if mu was made a github organization. The reasons for this are

1. i'm not interested in non-emacs commits or issues
2. i like to add all of the modules i use as git submodules in an elisp/ directory. This is not as nice to do when the elisp is in a subdirectory, and updates don't necessarily change anything for emacs.

I would probably also mirror mu's C source, but in a different directory.

When I run
$mu find £180
I get results mentioning 180 - preceded or not by the pound sign
$mu find £
returns
mu: no matches for search expression

From some point in time it become too slow. Especially annoying when you have speedbar open in background and it contantly refreshing maildir list. All emacs become blocking for several seconds even on my fresh SSD.

i belive its due to recent changes in mu4e~get-maildirs-1.

I have > than 60 folders in my ~/Maildir...

I can't delete message using the d key. Following is the error message (that doesn't say much, but you might have a idea :-):

Sure you want to execute marks on 1 message(s)? (y or n)  y
Error 70: path is /home/semente/Maildir/cur/1332954007.M772852P8849.magon,S=26291,W=26914:2,S, but maildir is /

Seems like Emacs 24 doesn't include the function user-error. C-h f user-error RET can't find it. Same in an Emacs instance started with -Q.

GNU Emacs 24.2.2 (x86_64-apple-darwin11.4.0, NS apple-appkit-1138.47) of 2012-08-28 on Moritzs-MacBook-Air

I am absolutely loving mu, but was curious as to what happened to the --format=json option? The manpage still says the option exists (probably needs to be updated) but it looks like commit 8e9bf21 gets rid of json, (according to the commit message)?

May I ask what the change is for? Obviously parsing the "plain" format is easy enough. In my particular case, I'm using python to invoke and parse the results, and the language makes it easy to translate from a json string to a python dict.

May we have json back? (Or, out of curiosity, why remove it?)

The documentation states that this var should define if mu4e-view buffers starts with the lines wrapped. That doesn't seem to work.

I tried grepping for mu4e-view-wrap-lines and the only occurence I fund is in mu4e-vars.el

Running mu4e (mu-0.9.8.5) in Emacs 24.2.1 on Ubuntu 12.04.

HTML message viewing is generally ok, but on certain messages (I believe all from the same source), I get the following error:

Debugger entered--Lisp error: (args-out-of-range "Page"" 6 5)
html2text-get-attr(573 717)
html2text-clean-anchor(573 717 727 731)
html2text-format-tags()
html2text()

It might be more of an html2text issue rather than mu4e, but I'm not sure at this point, and thought it might be worth reporting.

I can include the body text if it's helpful.

mu find v0.9.9-dev6 does not properly expand the ~ in the --linksdir parameter.
For example, the following command
mu find --muhome=/.mu-work --format=links --clearlinks --linksdir=/.localmail/search

will create a directory named
$HOME/.localmail/search/~/.localmail/search/

with correct search contents in it.
It would be correct if all search results were put into the maildir at ~/.localmail/search

I have one maildir with all my stuff in in it helpfully located at ~/mail.

I have a second collection of maildirs, in a directory called ~/work/mail.

For the purpose of getting mu to work right, I've created a bind mount of the later, at ~/mail/work which works like a dream. Mail gets indexed, I can read it in mu4e, life is good.

The problem is that to manipulate these folders (move messages, etc.) I have to actually type out full paths because the completion (i.e. after typing j or m) doesn't recurse one level.

I think fixing the completion stuff would make this 100% more usable for this case.

The exact error from make:

./.libs/libtestmucommon.a(mu-msg-crypto.o): In function `get_pkcs7_crypto_context':
/home/tom/repos/git/mu/lib/mu-msg-crypto.c:141: undefined reference to `GMIME_PKCS7_CONTEXT'

What I did:

% git pull origin master
% autoreconf -i
% ./configure && make

This probably applies as much to Sauron as it does to mu4e given they share very similar DBus documentation, but I spent a few hours tonight learning way more about DBus than I wanted to know due to a missing --type=method_call flag to dbus-send. Apparently the default is signal which is not listened for (or at least doesn't get caught under my configuration). This was further confused by the fact that running it in elisp

(dbus-call-method :session "org.gnu.Emacs" "/org/gnu/Emacs/Sauron" 
                  "org.gnu.Emacs.Sauron" "AddMsgEvent" "shell" 3 "foobar")

actually did produce the desired result. Maybe this is some sort of system default issue, but I don't think it could hurt to supply the proper argument value explicitly.

That is a feature that, as far I remember, is found in Gnus: when replying a message with some region marked (selected text), in compose mode buffer is yanked and quoted only that region.

Currently we can't integrate mu4e with the standard Emacs mail functionality because we cannot define a mail-user-agent for mu4e. The main thing holding us back is that we need a compose function that has the same interface as compose-mail.

Perhaps the best thing to do would be to define a mu4e-compose-mail with the compose-mail interface, and have the current compose function (which just grabs the information using mu4e-hdrs-* calls) call mu4e-compose-mail.

When I look at a message and want to come back to the preview view (list of the messages), I get this error when I press q:
error : mu4e-view-kill-buffer-and-window: Symbol's function definition is void: window-parent

I currently run offlineimap from cron and just after mu index, but I receive the following message when I have mu4e opened:

mu: mu_store_new_writable: xapian error 'Unable to get write lock on ~/.mu/xapian: already locked'

There is no way to let the user run mu index when the mu-server is running?

Thanks and thanks for mu/mu4e. It is everything I was looking for!

My Messages buffer is clogged with mu4e-title-face undefined errors; perhaps what was meant was mu4e-header-title-face?

This would be useful for many things: mu4e could display unread counts next to each bookmark/folder, users could use mu in their own scripts to display an unread count somewhere, etc. etc.

I get a failing test from running 'make check'. Carried on and did a 'make install' -- Haven't found any fault in the program so far.

commit e77abbe

TEST: test-mu-cmd... (pid=10356)
/mu-cmd/test-mu-index: OK
/mu-cmd/test-mu-find-empty-query: OK
/mu-cmd/test-mu-find-01: OK
/mu-cmd/test-mu-find-02: OK
/mu-cmd/test-mu-find-file: OK
/mu-cmd/test-mu-find-mime: OK
/mu-cmd/test-mu-find-links: OK
/mu-cmd/test-mu-find-text-in-rfc822: **
ERROR:test-mu-cmd.c:101:search: assertion failed (newlines_in_output(output) == expected): (0 == 1)
FAIL
GTester: last random seed: R02S1f2c20ce231db8a573c72696ab82c705
/bin/sh: line 5: 10340 Terminated gtester --verbose test-mu-query test-mu-contacts test-mu-cmd test-mu-cmd-cfind test-mu-threads
make[2]: *** [test] Error 143
make[2]: Leaving directory /home/mikkel/code/lisp/mu/mu/tests' make[1]: *** [test] Error 2 make[1]: Leaving directory/home/mikkel/code/lisp/mu/mu'
make: *** [test] Error 2

As of commit 96551f3dd to Org-Mode the function org-make-link is no longer defined and has been replaced with (concat ...).

There are two occurrences of this in org-mu4e.el, at lines 68 and 79.

If you want I can submit a pull request with the required change.

The manual specifies the format for adding new actions as
(add-to-list 'mu4e-headers-actions '("Number of recipients" ?n show-number-of-recipients) t)
whereas the code expects it to be a consed pair, like so:
(add-to-list 'mu4e-headers-actions '("Number of recipients" . show-number-of-recipients) t)

This is the case both in the 'actions' topic: http://www.djcbsoftware.nl/code/mu/mu4e/Actions.html
and the maintaining address-book with org-contacts: http://www.djcbsoftware.nl/code/mu/mu4e/Maintaining-an-address_002dbook-with-org_002dcontacts.html#Maintaining-an-address_002dbook-with-org_002dcontacts

Also, is there a way to insert email addresses in message-mode from org-contacts?

(Apologies for minor abuse of 'issues' thread with a query)

Would be create if you provide commands to label a message and show labelled messages.

How it could work:

1. labeling a message means add a header X-MU-Labels: label1, label, with comma, label2
2. a key binding to label a message (L) in mu4e and a similar command in CLI
3. a key binding to show messages with a certain label: type l so choose a label (as well a search parameter for mu CLI)
4. be able to create rules that work as labels (similar to mu4e-bookmarks), it could be in a configuration file (~/.mu-virtual-labels) and both mu4e mu CLI must interpret it

In the Maintaining an address book docs, you show this code snippet for registering an action:

(add-to-list 'mu4e-headers-actions
   '("org-contact-add" ?o mu4e-action-add-org-contact) t)

This results in a rather strange error, (111 mu4e-action-add-org-contact) is not a function. According to the Actions docs, the proper incantation is:

'("oorg-contact-add" . mu4e-action-add-org-contact) t)

Which indeed works as intended!

Show content of X-Label in message list/search result.

Faces for labels would be great also, the syntax could be similar to org-todo-keywork-faces:

(setq org-todo-keyword-faces
      '(("TESTING"  . (:foreground "#edec87" :weight bold))
        ("BLOCKED"  . (:foreground "#edec87" :weight bold))
        ("CANCELED" . (:foreground "#eeeeee" :weight bold))))

PS: I'm gonna use mu4e more often now, probably I will annoy you with lots of feature requests. Hope have time to help with these requests in a near future. :-)

The -n on line 47 in mu4e/Makefile.am causes problems on OS X:

@echo -n "(defconst mu4e-about \"" >> mu4e-about.el

It causes make to generate a mu4e-about.el starting with -n (defconst

Everything works fine when I manually remove the -n.

I don't think invoking an unknown bookmark should result in (error ...). It should invoke the standard bell and show a message in the minibuffer.

Here's the result of typing ba with a not being a configured bookmark:

Debugger entered--Lisp error: (error "Invalid shortcut 'a'")
  signal(error ("Invalid shortcut 'a'"))
  error("Invalid shortcut '%c'" 97)
  mu4e-get-bookmark-query(97)
  mu4e-ask-bookmark("Bookmark: ")
  mu4e-headers-search-bookmark()
  call-interactively(mu4e-headers-search-bookmark nil nil)

In the web page instructions said:

 $ tar xvfz mu-<version>.tar.gz  # use the specific version
 $ cd mu-<version>
 $./configure && make
 $ sudo make install

But en code of tag v0.9.8.4 I can't see the "configure" file.

Trying to build version mu-0.9.8.2 (as well as latest source) on ubuntu 10.04 I get the following error during `make check`` the reason I ran the check is after upgrading to 0.9.8.2 I was not seeing my sent mail saved into my sent folder.

Here is the output from the make check

TEST: test-mu-cmd-cfind... (pid=6425)
  /mu-cmd-cfind/test-mu-cfind-plain:                                   **
ERROR:test-mu-cmd-cfind.c:73:test_mu_cfind_plain: assertion failed (output == "Helmut Kröger [email protected]\n" "Mü [email protected]\n"): ("M\303\274 [email protected]\nHelmut Kr\303\266ger [email protected]\n" ==
274 [email protected]\n")
FAIL
GTester: last random seed: R02Sc482a8a6339c424d773c61f0676324dd
/bin/bash: line 5:  6230 Terminated              gtester --verbose test-mu-util test-mu-str test-mu-maildir test-mu-msg-fields test-mu-query test-mu-contacts test-mu-cmd test-mu-cmd-cfind test-mu-msg test-mu-r
u-date test-mu-flags
make[2]: *** [test] Error 143
make[2]: Leaving directory `/DIR/dev/mu/src/tests'
make[1]: *** [test] Error 2
make[1]: Leaving directory `/DIR/dev/mu/src'
make: *** [test] Error 2

emacs version 24.0.94 from source

mu configuration is complete.
-----------------------------
Xapian version                       : 1.2.7
GLib version                         : 2.24.1
GMime version                        : 2.4.14
GTK+ version                         : 2.20.1
GIO version                          : 2.24.1

If I type U on the mu4e-main window, offlineimap and mu run. Sometimes mu's indexer returns error code 1. I suppose this is related to Xapian locking, but unlike what is in the FAQ, I am not concurrently running mu4e and mu. Furthermore I can't give any more detailed information because mu deletes the mu4e-update buffer once mu fails. Any ideas what this problem could be? What information would be helpful for solving this problem?

Imap uses utf7 for unicode directory names (and in turn Maildirs). Ex:
まりもえお becomes &MH4wijCCMEgwSg-

However mu or at least mu4e shows just the &MH4wijCCMEgwSg- string.

Though I'm not really sure what's going on in message-mode, as best I can tell, after sending an e-mail (even with message-kill-buffer-on-exit set), a spare buffer containing my sent mail hangs around, I think due to copying the file out to the FCC directory. Could there be a mu4e option that would enable automatic killing/cleanup of those FCC buffers? (Would it be as simple as calling (kill-buffer) at the end of the message-fcc-handler-function in mu4e~setup-fcc-maybe?)

Thanks!

I currently use mu4e for most of my email needs. It is great and makes the very dull task of dealing with messages so much more enjoyable, or at least tolerable. The one thing keeping me from fully adopting mu4e is the lack of support for encrypted messages. One of my email accounts has all emails automatically encrypted, so this is a big deal for me. Notmuch currently does a decent job with this, even better than Gnus, but neither are perfect, sometimes requiring manual decryption. Adding this feature would make me very happy. I would be willing to help if at all possible.

Thanks for considering this request.

Hi,

When I jump to a mail folder for the first time, it is very slow to show the list of messages.

My inbox has 56 unread messages in it currently and it takes 32 seconds to show anything, with no indication that its actually doing something.

To reproduce, start emacs, start mu4e, press j followed by to get the default folder (which is my inbox)

wait 32 seconds, then mu4e shows emails.

I tried setting debug on quit, but when i pressed C-g it did not seem to create a debug buffer for some reason.

Jumping to inbox again is instant. quit emacs and restart, and then I have to wait 30 seconds gain for mails to show.

jumping to any other folder also seems to be affected.

Jason

The fact that the display in the fringe listing your temporary marks for the "mark as read" mark is "R" is confusing -- it makes me keep wanting to hit 'R' to set the read mark when I really should be hitting 'r'. The move mark, for example, is 'm', not 'M'. Any reason not to change the display to lowercase?

(Not that I think you need a patch for something you wrote by yourself from scratch, but mu4e-mark.el, line 106, or line 104 in trunk.)

A good PGP/MIME support is crucial for every mail client. It would be very cool if mu(4e) would support this. I saw that you already started implementing it, so I thought it would be ok to open a feature request for people to comment on this.

I am willing to donate for this feature, as I don't have the knowledge of C/C++ and PGP to implement this myself.

Wow. mu4e looks amazing on those screenshots! Although I am somewhat happy with my wanderlust setup (thanks to emacs-fu!) there are some things that could be better. And it looks like mu4e does them better. I don't have enough time to test it right now, but is it possible to store mu4e emails in orgmode? And does it work with offlineimap? If so, I might start testing it in a week or two.

BTW, I like your kind of 'vision' (or simply your way of thinking) when it comes to emacs. Sauron looks like a fantastic idea, too.

Despite always working in the past (and having org), my most recent attempt fails to compile org-mu4e.el, citing no org library. Removing it from the dependencies in mu4e/Makefile did successfully build.

When telling mu4e to apply all changes when closing a header window via

(setq mu4e-headers-leave-behavior 'apply)

it doesn't do anything.

I noticed the mu4e-headers-leave-behavior is defined via defcustom. Could this cause incompatibilities with my way of setting the value? (I really don't want to use custom)

Functions like mu4e-view-mark-for-trash set the mark for the message at point in the headers buffer, not for the message currently viewed in the message buffer. This behavior is a bit confusing and can lead to unwanted and unnoticeable marking of messages, e.g. if you (accidentaly) scrolled the headers buffer while reading the message (happened to me.. ;) ). Maybe mu4e-mark-at-point could be given a docid argument. Then, either a (optional) warning could pop up if the docid does not match the docid at point, or the docid argument is preferred over the docid at point (I would prefer the latter. Can't think of a reason why anyone would prefer the current behavior).

The Emacs23 work-around that you have bound the keymap "q" (in mu4e-view.el, maybe others) is killing all of my windows when I try to get out of viewing a message:

;; emacs23 hack: trial and error
(condition-case nil
  (kill-buffer-and-window)
  (kill-buffer))

As a result, I do properly see mu4e-main after 'quitting' the view or header mode, but it kills my 5-6 other open windows in the process. I am running Emacs 23.4.

I was going through my inbox today marking stuff for archival (d) and after about 50 or so marks, Emacs started consuming all available CPU/memory and thrashing the disk (presumably due to swapping). I've been able to reproduce this pretty reliably; the system starts becoming sluggish, then entirely unresponsive as the number of marks increase. Executing all marked actions puts things back to normal.