This is the locale that should be used for computer interoperability
when doing date and time parsing and formatting.
Without this, on systems which don't have the US locale set, the parsing
returns 0.
With the eager evaluation, you run the risk of checking the capabilities
store before any connection to the server may have been opened.
Therefore, the capabilities uid_store will be empty and it will fall
back to poll_with_examine even if the server might have support for
idle.
A lazy stateful iterator for line breaking text. Useful for very long text where you don't want to linebreak it completely before user requests specific lines.
When inserting an envelope in a thread and its Message-ID already exists
with an associated envelope, overwrite the association if the previous
associated envelope is from a foreign mailbox and current envelope is
not. This happens when mail from a sent folder has been inserted in eg
your INBOX, but somehow INBOX has a copy of your own message as well.
This can happen when mailing lists that send you copies of your own
posts.
The problem with this was that in IMAP your mailing list copy was unseen
and you could not mark it seen because the thread only knew about your
Sent mailbox copy.
In examine_updates() which is periodically executed in the IMAP watch
thread, the mailbox's contents get fetched regardless if the user has
fetched the mailbox before. That means eg a large mailbox that was
unused by the user might perform a large fetch all of a sudden without
the user's knowledge/touch.
Add `warm` property in ImapMailbox that states whether the mailbox has
been loaded before in current execution.
Closes#88 IMAP: don't examine unloaded mailboxes for updates, just for message count stats
In some cases when handling new server events, the entire body message
was unnecessarily fetched.
Closes#87 IMAP: don't fetch RFC822 except when requested
Previous commit changed bincode deserializes in maildir and sqlite3.rs
from bincode::deserialize_from to using bincode::config::DefaultOptions
and bincode::Options trait's method deserialize_from.
However, these two different deserializes use a different default
settings: https://github.com/servo/bincode/issues/348
Specifically, varint encoding for integers is the default for
DefaultOptions but not when using bincode::{de,}serialize_* functions.
That means that serialized structs were not able to be deserialized.
This commit makes all {de,}serializations use the DefaultOptions
interface rather than the top level functions.
If struct memory layout changes, bincode deserialize fails with memory
allocation error of an obscene amount of bytes. Set upper limit to
deserialized bytes in each place deserialize happens.
Meli currently uses the .well-known/jmap URL and the RFC8620 requires that any redirects are followed (https://tools.ietf.org/html/rfc8620#section-2.2). This small change allows redirects to happen.
The widths of subslices of a line are calculated in each call to
`binary_search_by` when reflowing long lines. This can be done in Ologn
queries with a segment tree.
Manos Pitsidianakis
Zisu Andrei
Rudi Horn