On Mon, 12 May 2014, David Edmondson <dme@dme.org> wrote: > On Sun, May 11 2014, Mark Walters wrote: >> This adds a function that highlights searches which have newly arrived >> messages (ones which arrived since the previous refresh of >> notmuch-hello). It does that by getting a full list of matching >> message ids and checking whether any new messages have appeared. Thus >> it will be slow on large saved searches. >> --- >> emacs/notmuch-hello.el | 21 +++++++++++++++++++++ >> 1 file changed, 21 insertions(+) >> >> diff --git a/emacs/notmuch-hello.el b/emacs/notmuch-hello.el >> index 28e16be..265b879 100644 >> --- a/emacs/notmuch-hello.el >> +++ b/emacs/notmuch-hello.el >> @@ -96,6 +96,8 @@ (define-widget 'notmuch-saved-search-plist 'list >> (const :tag "Default (messages)" nil) >> (function-item :tag "messages/threads" >> notmuch-hello-display-count-threads-and-messages) >> + (function-item :tag "highlight newly arrived messages" >> + notmuch-hello-display-new-messages) >> function))))) >> >> (defcustom notmuch-saved-searches '((:name "inbox" :query "tag:inbox") >> @@ -507,6 +509,25 @@ (defun notmuch-hello-display-count-threads-and-messages (&rest args) >> (notmuch-hello-nice-number threads)))) >> (plist-put current :count display))) >> >> +(defun notmuch-hello-display-new-messages (&rest args) >> + "Highlight saved searches which have new messages. >> + >> +New messages are ones which have appeared since the last time >> +notmuch hello was refreshed. This will be slow on large queries." >> + (let* ((current (plist-get args :current)) >> + (old (plist-get args :old)) >> + (query (plist-get current :query)) >> + (name (plist-get current :name)) >> + (new-list (notmuch-call-notmuch-sexp "search" >> + "--output=messages" >> + "--format=sexp" >> + query)) >> + (old-list (plist-get old :message-list))) >> + (unless (subsetp new-list old-list :test 'equal) >> + (plist-put current :name (propertize name 'face '(:foreground "red")))) >> + (plist-put current :count (notmuch-hello-nice-number (length new-list))) >> + (plist-put current :message-list new-list))) > > I wonder if it would be better to store a hash value derived from the > message list rather than the list itself. The list could be large. My view was that the list was OK as a user would be foolish to set this for any large queries anyway (as the query itself would be too slow). Also, a hash itself wouldn't get the effect I want as I do want to be to test "subset" for the two lists (as I don't care about messages no longer matching, only new messages that do match: think inbox type things) I could store the hash of the message-id. That might save a substantial chunk on storage (eg 40 byte message if to 4byte hash or something) More generally, it is not clear to me whether we want to make it easy for a user to cause notmuch-hello to be very slow. Perhaps we need a timing for each display-function. If any one takes over 0.5s then we can report it to the user! Best wishes Mark > >> + >> (defun notmuch-hello-batch-message-count (elem-plist options) >> "Update the message count for a saved search. >> >> -- >> 1.7.10.4 >> >> _______________________________________________ >> notmuch mailing list >> notmuch@notmuchmail.org >> http://notmuchmail.org/mailman/listinfo/notmuch