Given a range of entries that occupy a page (eg [0, 50] for a page of 50
rows high) get the max entry width for this column by using maximum
range queries with segment trees.
- set_subject checked if last byte was control character instead of last
character. Characters can be multi-byte, duh.
- email::parser::date didn't provide for Date values that had -0000
instead of +0000 (that's a chrono requirement/bug)
- melib/imap: accept quoted strings with escaped quotes in
protocol_parser
- ui/accounts: return unavailabity correctly if folder's worker slot is
empty instead of judging only by its vacancy
- ui/MailView: set view as not dirty if envelope loading from backend
fails so that it stops requesting it in every subsequent redraw
Eventually after loading potential usage values from configuration,
backends will be able to change the usage values themselves. IMAP and
JMAP have the ability to set Mailbox roles (IMAP needs LIST-SPECIAL
extension
Add tags() method that returns Option<Arc<RwLock<BTreeMap<u64, String>>>>.
The BTreeMap holds available tags in a mail backend and uses the tag's
hash as key.
The method returns an Option because not all backends may support
tagging.
Change folders() signature:
- fn folders(&self) -> FnvHashMap<FolderHash, Folder>;
+ fn folders(&self) -> Result<FnvHashMap<FolderHash, Folder>>;
Imap may not be online, therefore we need the ability to return an
error.
- Parse List-Post value like List-Unsubscribe: comma separated angle bracket limited list of <mailto:> or <url> values
- Check if List-Archive value is angle bracket delimited
fetch BODYSTRUCTURE along with ENVELOPE from server and set
has_attachments based on the MIME structure of the envelope.
Notes: BODYSTRUCTURE returns the MIME structure of the envelope without
the data, so if it includes a multipart/mixed it *should* have
attachments.
ENVELOPE returns basic headers of the message like Sender, Subject, Date
etc.
thread_group property of ThreadNode doesn't yet reflect the actual root
ThreadNode (the root of the thread, that is). So find the root manually
instead.
Make startup methods return Results so that the main binary can exit
cleanly instead of using std::process::exit from arbitrary positions,
which exits the process immediately and doesn't run destructors.
