New User Guide

Welcome to Sup! Here's how to get started.

First, try running sup. Since this is your first time, you'll be confronted with a mostly blank screen, and a notice at the bottom that you have no new messages. That's because Sup hasn't loaded anything into its index yet, and has no idea where to look for them anyways.

If you want to play around a little at this point, you can press 'b' to cycle between buffers, ';' to get a list of the open buffers, and 'x' to kill a buffer. There's probably not too much interesting there, but there's a log buffer with some cryptic messages. You can also press '?' at any point to get a list of keyboard commands, but in the absence of any email, these will be mostly useless. When you get bored, press 'q' to quit.

To use Sup for email, we need to load messages into the index. The index is where Sup stores all message state (e.g. read or unread, any message labels), and all information necessary for searching and for threading messages. Sup only knows about messages in its index.

We can add messages to the index by telling Sup about the "source" where the messages reside. Sources are things like mbox folders, and maildir directories. Sup doesn't duplicate the actual message content in the index; it only stores whatever information is necessary for searching, threading and labelling. So when you search for messages or view your inbox, Sup talks only to the index (stored locally on disk). When you view a thread, Sup requests the full content of all the messages from the source.

The easiest way to set up all your sources is to run sup-config. This will interactively walk you through some basic configuration, prompt you for all the sources you need, and optionally import messages from them. Sup-config uses two other tools, sup-add and sup-sync, to load messages into the index. In the future you may make use of these tools directly (see below).

Once you've run sup-config, you're ready to run sup. You should see the most recent unarchived messages appear in your inbox. Congratulations, you've got Sup working!

If you're coming from the world of traditional MUAs, there are a couple differences you should be aware of at this point. First, Sup has no folders. Instead, you organize and find messages by a combination of search and labels (known as "tags" everywhere else in the world). Search and labels are an integral part of Sup because in Sup, rather than viewing the contents of a folder, you view the results of a search. I mentioned above that your inbox is, by definition, the set of all messages that aren't archived. This means that your inbox is nothing more than the result of the search for all messages with the label "inbox". (It's actually slightly more complicated---we also omit messages marked as killed, deleted or spam.)

You could replicate the folder paradigm easily under this scheme, by giving each message exactly one label and only viewing the results of simple searches for those labels. But you'd quickly find out that life can be easier than that if you just trust the search engine, and use labels judiciously for things that are too hard to find with search. The idea is that a labeling system that allows arbitrary, user-defined labels, supplemented by a quick and easy-to-access search mechanism provides all the functionality that folders does, plus much more, at a far lower cost to the user.

Now let's take a look at your inbox. You'll see that Sup groups messages together into threads: each line in the inbox is a thread, and the number in parentheses is the number of messages in that thread. (If there's no number, there's just one message in the thread.) In Sup, most operations are on threads, not individual messages. The idea is that you rarely want to operate on a message independent of its context. You typically want to view, archive, kill, or label all the messages in a thread at one time.

Use the up and down arrows to highlight a thread. ('j' and 'k' do the same thing, and 'J' and 'K' will scroll the whole window. Even the left and right arrow keys work.) By default, Sup only loads as many threads as it takes to fill the window; if you'd like to load more, press 'M'. You can hit tab to cycle between only threads with new messages.

Highlight a thread and press enter to view it. You'll notice that all messages in the thread are displayed together, laid out graphically by their relationship to each other (replies are nested under parents). By default, only the new messages in a thread are expanded, and the others are hidden. You can toggle an individual message's state by highlighting a green line and pressing enter. You can use 'E' to expand or collapse all messages or 'N' to expand only the new messages. You'll also notice that Sup hides quoted text and signatures. If you highlight a particular hidden chunk, you can press enter to expand it, or you can press 'o' to toggle every hidden chunk in a particular message.

Other useful keyboard commands when viewing a thread are: 'n' and 'p' to jump to the next and previous open messages, 'h' to toggle the detailed headers for the current message, and enter to expand or collapse the current message (when it's on a text region). Enter and 'n' in combination are useful for scanning through a thread---press enter to close the current message and jump to the next open one, and 'n' to keep it open and jump. If the buffer is misaligned with a message, you can press 'z' to highlight it.

This is a lot to remember, but you can always hit '?' to see the full list of keyboard commands at any point. There's a lot of useful stuff in there---once you learn some, try out some of the others!

Now press 'x' to kill the thread view buffer. You should see the inbox again. If you don't, you can cycle through the buffers by pressing 'b', or you can press ';' to see a list of all buffers and simply select the inbox.

There are many operations you can perform on threads beyond viewing them. To archive a thread, press 'a'. The thread will disappear from your inbox, but will still appear in search results. If someone replies an archived thread, it will reappear in your inbox. To kill a thread, press '&'. Killed threads will never come back to your inbox, even if people reply, but will still be searchable. (This is useful for those interminable threads that you really have no immediate interest in, but which seem to pop up on every mailing list.)

If a thread is spam, press 'S'. It will disappear and won't come back. It won't even appear in search results, unless you explicitly search for spam.

You can star a thread by pressing '*'. Starred threads are displayed with a little yellow asterisk next to them, but otherwise have no special semantics. But you can also search for them easily---we'll see how in a moment.

To edit the labels for (all the messages in) a thread, press 'l'. Type in the labels as a sequence of space-separated words. To cancel the input, press Ctrl-G.

Many of these operations can be applied to a group of threads. Press 't' to tag a thread. Tag a couple, then press '=' to apply the next command to the set of threads. '=t', of course, will untag all tagged messages.

Ok, let's try using labels and search. Press 'L' to do a quick label search. You'll be prompted for a label; simply hit enter to bring up scrollable list of all the labels you've ever used, along with some special labels (Draft, Starred, Sent, Spam, etc.). Highlight a label and press enter to view all the messages with that label.

What you just did was actually a specific search. For a general search, press '' (backslash---forward slash is used for in-buffer search, following console conventions). Now type in your query (again, Ctrl-G to cancel at any point.) You can just type in arbitrary text, which will be matched on a per-word basis against the bodies of all email in the index, or you can make use of the full Xapian query syntax http://xapian.org/docs/queryparser.html:

• Phrasal queries using double-quotes, e.g.: "three contiguous words"
• Queries against a particular field using :, e.g.: label:ruby-talk, or from:[email protected]. (Fields include: body, from, to, and subject.)
• Force non-occurrence by -, e.g. -body:"hot soup".
• If you have the chronic gem installed, date queries like "before:today", "on:today", "after:yesterday", "after:(2 days ago)" (parentheses required for multi-word descriptions).

You can combine those all together. For example:

 label:ruby-talk subject:[ANN] -rails on:today

Play around with the search, and see the Xapian documentation for details on more sophisticated queries (date ranges, "within n words", etc.)

At this point, you're well on your way to figuring out all the cool things Sup can do. By repeated application of the '?' key, see if you can figure out how to:

• List some recent contacts
• Easily search for all mail from a recent contact
• Easily search for all mail from several recent contacts
• Add someone to your address book
• Postpone a message (i.e., save a draft)
• Quickly re-edit a just-saved draft message
• View the raw header of a message
• Star an individual message, not just a thread

There's one last thing to be aware of when using Sup: how it interacts with other email programs. As I described above, Sup stores data about messages in the index, but doesn't duplicate the message contents themselves. The messages remain on the source. If the index and the source every fall out of sync, e.g. due to another email client modifying the source, then Sup will be unable to operate on that source. For example, for mbox files, Sup stores a byte offset into the file for each message. If a message deleted from that file by another client, or even marked as read (yeah, mbox sucks), all succeeding offsets will be wrong.

That's the bad news. The good news is that Sup is pretty good at being able to detect this type of situation, and fixing it is just a matter of running sup-sync --changed on the source. Sup will even tell you how to invoke sup-sync when it detects a problem. This is a complication you will almost certainly run in to if you use both Sup and another MUA on the same source, so it's good to be aware of it.

Have fun, and email [email protected] if you have any problems!

Appendix A: sup-add and sup-sync

Instead of using sup-config to add a new source, you can manually run sup-add with a URI pointing to it. The URI should be of the form:

• mbox://path/to/a/filename, for an mbox file on disk.
• maildir://path/to/a/filename, for a maildir directory on disk.

Before you add the source, you need make three decisions. The first is whether you want Sup to regularly poll this source for new messages. By default it will, but if this is a source that will never have new messages, you can specify --unusual. Sup polls only "usual" sources when checking for new mail (unless you manually invoke sup-sync).

The second is whether you want messages from the source to be automatically archived. An archived message will not show up in your inbox, but will be found when you search. (Your inbox in Sup is, by definition, the set of all all non-archived messages). Specify --archive to automatically archive all messages from the source. This is useful for sources that contain, for example, high-traffic mailing lists that you don't want polluting your inbox.

The final decision is whether you want any labels automatically applied to messages from this source. You can use --labels to do this.

Now that you've added the source, let's import all the current messages from it, by running sup-sync with the source URI. You can specify --archive to automatically archive all messages in this import; typically you'll want to specify this for every source you import except your actual inbox. You can also specify --read to mark all imported messages as read; the default is to preserve the read/unread status from the source.

Sup-sync will now load all the messages from the source into the index. Depending on the size of the source, this may take a while. Don't panic! It's a one-time process.

Appendix B: Automatically labeling incoming email

One option is to filter incoming email into different sources with something like procmail, and have each of these sources auto-apply labels by using sup-add --labels.

But the better option is to learn Ruby and write a before-add hook. This will allow you to apply labels based on whatever crazy logic you can come up with. See Hooks for more.

Appendix C: Reading blogs with Sup

Really, blog posts should be read like emails are read---you should be able to mark them as unread, flag them, label them, etc. Use rss2email to transform RSS feeds into emails, direct them all into a source, and add that source to Sup. Voila!