Re: nevermore

[Trevor Jim <tjim@mac.com>]

[-- Attachment #1: Type: text/plain, Size: 1626 bytes --]

If you would like to incorporate things into the main branch, that would be great. Here's what I think the status is:

The asynchronous mail address harvesting works pretty well. Much better than having to install python etc. It would be better to also update the addresses when sending mail.

I imagine many people want to use something other than company-mode, though. Probably needs testing for other completion systems.

The incremental search works well for me, but I think it needs testing on larger maildirs. Notmuch as it exists today does not seem to have incremental search in mind. For example, I have to kill and start a new process each time the search term changes (potentially on each keystroke).

My async parsing is noticeably faster than notmuch.el, so you might want to incorporate some of its ideas, regardless of whether you take incremental search.

One complication of the incremental search code is that I like to query based on messages rather than threads. Both are actually useful, but I'm happier with messages. (For one thing, it should be much faster.) Notmuch, of course, is thread-focused.

For message-based search, --output=summary is not appropriate, even if the query is a message-id; the result includes, for example, tags that appear only in other messages in the thread. So I have been forced to use notmuch show --body=false instead. This precludes other useful options of notmuch search, e.g., --sort. Anyway, the incremental search code has some complications for this. I'd love for notmuch search to have a --output=message-summary option to avoid this.

-Trevor


[-- Attachment #2.1: Type: text/html, Size: 1692 bytes --]