Fix the things that are causing the most pain to new users ---------------------------------------------------------- 1. A new import is tagging all messages as "inbox" -- total pain Emacs interface (notmuch.el) ---------------------------- Add notmuch-bcc and notmuch-cc for setting default Bcc and Cc values, (should affect the message-setup-hook). Switch the notmuch-search view to use "notmuch search --format=json" to fix large classes of bugs regarding poorly-escaped output and lame regular expressions. (The most recently found, unfixed example is the sender's name containing ';' which causes emacs to drop a search result.) This may require removing the outer array from the current "notmuch search --format=json" results. Add a global keybinding table for notmuch, and then view-specific tables that add to it. Add a '|' binding from the search view. Add support for choosing from one of the user's configured email addresses for the From line. Make 'notmuch-show-pipe-message have a private history. Add support for a delete keybinding that adds a "deleted" tag to the current message/thread and make searches not return deleted messages by default, (unless the user asks explicitly for deleted messages in the search query). Add support to "mute" a thread (add a "muted" tag and then don't display threads in searches by default where any message of the thread has the "muted" tag). Make '=' count from the end rather than from the beginning if more than half-way through the buffer. Fix to automatically wrap long headers (for RFC compliance) before sending. This should probably just be fixed in message-mode itself, (but perhaps we can have a notmuch-message-mode that layers this on top). Stop hiding the headers so much in the thread-view mode. Allow opening a message in thread-view mode by clicking on either line. Automatically open a message when navigating to it with N or P. Change 'a' command in thread-view mode to only archive open messages. notmuch command-line tool ------------------------- Add support to "notmuch search" and "notmuch show" to allow for listing of duplicate messages, (distinct filenames with the same Message-ID). I'm not sure what the option should be named. Perhaps --with-duplicates ? Replace "notmuch reply" with "notmuch compose --reply ". This would enable a plain "notmuch compose" to be used to construct an initial message, (which would then have the properly configured name and email address in the From: line. We could also then easily support "notmuch compose --from " to support getting at alternate email addresses. Implement "notmuch search --exclude-threads=" to allow for excluding muted threads, (and any other negative, thread-based filtering that the user wants to do). Fix "notmuch show" so that the UI doesn't fail to show a thread that is visible in a search buffer, but happens to no longer match the current search. (Perhaps add a --matching= option (or similar) to "notmuch show".) For now, this is being worked around in the emacs interface by noticing that "notmuch show" returns nothing and re-rerunning the command without the extra arguments. Add a "--format" option to "notmuch search", (something printf-like for selecting what gets printed). Give "notmuch restore" some progress indicator. Fix "notmuch restore" to operate in a single pass much like "notmuch dump" does, rather than doing N searches into the database, each matching 1/N messages. Allow configuration for filename patterns that should be ignored when indexing. Fix to avoid this ugly message: (process:17197): gmime-CRITICAL **: g_mime_message_get_mime_part: assertion `GMIME_IS_MESSAGE (message)' failed Warning: Not indexing empty mime part. This probably means adding a test case to generate that message, filing an upstream bug against GMime, and then silencing the notmuch-generated portion of the warning (so that once GMime is fixed, this is all silent). Simplify notmuch-reply to simply print the headers (we have the original values) rather than calling GMime (which encodes) and adding the confusing gmime-filter-headers.c code (which decodes). Properly handle replying to multiple messages. Currently, the JSON reply format only supports a single message, but the default reply format accepts searches returning multiple messages. The expected behavior of replying to multiple messages is not obvious, and there are multiple ideas that might make sense. Some consensus needs to be reached on this issue, and then both reply formats should be updated to be consistent. Return docid-based queries in thread search results, instead of message-id-based queries. These are much more compact and much faster to later query. This will require support from the library, both for retrieving docids (probably as an opaque "get a query that identifies this message") and for docid queries. notmuch library --------------- Add support for custom flag<->tag mappings. In the notmuch configuration file this could be [maildir] synchronize_flags = R:replied; D*:deleted; S:~unread; In the library interface this could be implemented with an array of structures to define the mapping (flag character, tag name, inverse-sense bit (~ above), and tag-when-any-file-flagged vs. tag-when-all-files-flagged (* above)). Add an interface to accept a "key" and a byte stream, rather than a filename. Improve syntax for date ranges queries. date:expr should be interpreted as date:expr..expr so that, for example, "date:2013-01-22" would cover the whole of the specified day (currently that's not even recognized as a date range expression). It might be nice to be able to use things like "since:2013-01-22" and "until:2013-01-22" as synonyms to "date:2013-01-22.." and "date:..2013-01-22", respectively. To do any of this we're probably going to need to break down and write our own parser for the query string rather than using Xapian's QueryParser class. Make failure to read a file (such as a permissions problem) a warning rather than an error (should be similar to the existing warning for a non-mail file). Fix to use the *last* Message-ID header if multiple such headers are encountered, (I noticed this is one thing that kept me from seeing the same message-ID values as sup). Add support for configuring "virtual tags" which are a tuple of (tag-name, search-specification). The database is responsible for ensuring that the virtual tag is always consistent. Indicate to the user if two files with the same message ID have content that is actually different in some interesting way. Perhaps notmuch initially sees all changes as interesting, and quickly learns from the user which changes are not interesting (such as the very common mailing-list footer). Fix notmuch_query_count_messages to share code with notmuch_query_search_messages rather than duplicating code. (And consider renaming it as well.) Provide a mechanism for doing automatic address completion based on notmuch searches. Here was one proposal made in IRC: I guess all it would really have to be would be a way to configure a series of searches to try in turn, (presenting ambiguities at a given single level, and advancing to the next level only if one level returned no matches). So then I might have a series that looks like this: notmuch search --output=address_from tag:address_book_alias notmuch search --output=address_to tag:sent notmuch search --output=address_from I think I might like that quite a bit. And then we have a story for an address book for non-emacs users. Provide a ~me Xapian synonym for all of the user's configured email addresses. Add symbol hiding so that we don't risk leaking any private symbols into the shared-library interface. Audit all libnotmuch entry points to ensure that all Xapian calls are wrapped in a try/catch block. Fix the threading of a message that has a References: header but no In-Reply-To: header (see id:"87lixxnxpb.fsf@yoom.home.cworth.org"). Search syntax ------------- Implement support for "tag:*" to expand to all tags. Fix "notmuch search to:" to be less confusing. Many users expect this to search for all messages with a To: header, but it instead searches for all messages with the word "to". If we don't provide the first behavior, perhaps we should exit on an error when a configured prefix is provided with no value? Support "*" in all cases and not just as a special case. That is, "* " should also work, as well as "* and tag:inbox". Implement a syntax for requesting set-theoertic operations on results of multiple searches. For example, I would like to do: "tag:inbox" SET-SUBTRACT "tag:muted" as well as: "tag:notmuch and " SET-INTERSECT "tag:notmuch and not (tag:merged or tag:postponed)" See id:3wdpr282yz2.fsf@testarossa.amd.com for more details on the use cases of the above. Database changes ---------------- Store a reference term for every message-id that appears in References. We just started doing this for newly-added documents, but at the next convenient database-schema upgrade, we should go back and fix old messages to be consistent. Start indexing the List-Id header, (and re-index this header for existing messages at the next database upgrade). Add support for the user to specify custom headers to be indexed (and re-index these for existing messages at the next database upgrade). Save filenames for files detected as "not an email file" in the database. This would allow for two things: 1. Optimizing "notmuch new" to not have to look at these files again (since they are potentially large so the detection could be potentially slow). 2. A "notmuch search" syntax could be added to allow the user to find these files, (and perhaps delete them or move them away as appropriate). Fix filesystem/notmuch-new race condition by not updating database mtime for a directory if it is the same as the current mtime. Test suite ---------- Achieve 100% test coverage with the test suite. General ------- Audit everything for dealing with out-of-memory (and drop xutil.c). Investigate why the notmuch database is slightly larger than the sup database for the same corpus of email. Makefile should print message teaching user about LD_LIBRARY_PATH (or similar) if libdir is not set to a directory examined by ldconfig.