Currently there is a race between transports
to upload sync messages and delete them from `imap_send` table.
Sometimes mulitple transports upload the same message
and sometimes only some of them "win".
With this change only the primary transport
will upload the sync message.
For multiple transports we will need to run
multiple IMAP clients in parallel.
UID validity change detected by one IMAP client
should not result in UID resync
for another IMAP client.
The motivation is to reduce code complexity, get rid of the extra IMAP connection and cases when
messages are added to chats by Inbox and Sentbox loops in parallel which leads to various message
sorting bugs, particularly to outgoing messages breaking sorting of incoming ones which are fetched
later, but may have a smaller "Date".
We already have both rand 0.8 and rand 0.9
in our dependency tree.
We still need rand 0.8 because
public APIs of rPGP 0.17.0 and Iroh 0.35.0
use rand 0.8 types in public APIs,
so it is imported as rand_old.
We do not try to delete resent messages
anymore. Previously resent messages
were distinguised by having duplicate Message-ID,
but future Date, but now we need to download
the message before we even see the Date.
We now move the message to the destination folder
but do not fetch it.
It may not be a good idea to delete
the duplicate in multi-device setups anyway,
because the device which has a message
may delete the duplicate of a message
the other device missed.
To avoid triggering IMAP busy move loop
described in the comments
we now only move the messages
from INBOX and Spam folders.
This way, the statistics / self-reporting bot will be made into an
opt-in regular sending of statistics, where you enable the setting once
and then they will be sent automatically. The statistics will be sent to
a bot, so that the user can see exactly which data is being sent, and
how often. The chat will be archived and muted by default, so that it
doesn't disturb the user.
The collected statistics will focus on the public-key-verification that
is performed while scanning a QR code. Later on, we can add more
statistics to collect.
**Context:**
_This is just to give a rough idea; I would need to write a lot more
than a few paragraphs in order to fully explain all the context here_.
End-to-end encrypted messengers are generally susceptible to MitM
attacks. In order to mitigate against this, messengers offer some way of
verifying the chat partner's public key. However, numerous studies found
that most popular messengers implement this public-key-verification in a
way that is not understood by users, and therefore ineffective - [a 2021
"State of Knowledge" paper
concludes:](https://dl.acm.org/doi/pdf/10.1145/3558482.3581773)
> Based on our evaluation, we have determined that all current E2EE
apps, particularly when operating in opportunistic E2EE mode, are
incapable of repelling active man-in-the-middle (MitM) attacks. In
addition, we find that none of the current E2EE apps provide better and
more usable [public key verification] ceremonies, resulting in insecure
E2EE communications against active MitM attacks.
This is why Delta Chat tries to go a different route: When the user
scans a QR code (regardless of whether the QR code creates a 1:1 chat,
invites to a group, or subscribes to a broadcast channel), a
public-key-verification is performed in the background, without the user
even having to know about this.
The statistics collected here are supposed to tell us whether Delta Chat
succeeds to nudge the users into using QR codes in a way that is secure
against MitM attacks.
**Plan for statistics-sending:**
- [x] Get this PR reviewed and merged (but don't make it available in
the UI yet; if Android wants to make a release in the meantime, I will
create a PR that removes the option there)
- [x] Set the interval to 1 week again (right now, it's 1 minute for
testing)
- [ ] Write something for people who are interested in what exactly we
count, and link to it (see `TODO[blog post]` in the code)
- [ ] Prepare a short survey for participants
- [ ] Fine-tune the texts at
https://github.com/deltachat/deltachat-android/pull/3794, and get it
reviewed and merged
- [ ] After the next release, ask people to enable the
statistics-sending
Follow-up to https://github.com/chatmail/core/pull/7125: We now have a
mix of non-async (parking_lot) and async (tokio) Mutexes used for the
connectivity. We can just use non-async Mutexes, because we don't
attempt to hold them over an await point. I also tested that we get a
compiler error if we do try to hold one over an await point (rather than
just deadlocking/blocking the executor on runtime).
Not 100% sure about using the parking_lot rather than std Mutex, because
since https://github.com/rust-lang/rust/issues/93740, parking_lot
doesn't have a lot of advantages anymore. But as long as iroh depends on
it, we might as well use it ourselves.
`get_connectivity()` is expected to return immediately, not when the scheduler finishes updating its
state in `start_io()/stop_io()/pause_io()`, otherwise it causes app non-responsiveness.
Instead of read-locking `SchedulerState::inner`, store the `ConnectivityStore` collection in
`Context` and fetch it from there in `get_connectivity()`. Update it every time we release a write
lock on `SchedulerState::inner`.
Save:
- (old contact id) -> (new contact id) mapping.
- The message id starting from which all messages are already migrated.
Run the migration from `housekeeping()` for at least 500 ms and for >= 1000 messages per run.
Why:
- With IMAP APPEND we can upload messages directly to the DeltaChat folder (for non-chatmail
accounts).
- We can set the `\Seen` flag immediately so that if the user has other MUA, it doesn't alert about
a new message if it's just a sync message (there were several such reports on the support
forum). Though this also isn't useful for chatmail.
- We don't need SMTP envelope and overall remove some overhead on processing sync messages.
This fixes things for Gmail f.e. Before, `Imap::fetch_move_delete()` was called before looking for
Trash and returned an error because of that failing the whole `fetch_idle()` which prevented
configuring Trash in turn.
- Always emit `ContactsChanged` from `contact::update_last_seen()` if a contact was seen recently
just for simplicity and symmetry with `RecentlySeenLoop::run()` which also may emit several events
for single contact.
- Fix sleep time calculation in `RecentlySeenLoop::run()` -- `now` must be updated on every
iteration, before the initial value was used every time which led to progressively long sleeps.
Connection establishment now happens only in one place in each IMAP loop.
Now all connection establishment happens in one place
and is limited by the ratelimit.
Backoff was removed from fake_idle
as it does not establish connections anymore.
If connection fails, fake_idle will return an error.
We then drop the connection and get back to the beginning of IMAP
loop.
Backoff may be still nice to have to delay retries
in case of constant connection failures
so we don't immediately hit ratelimit if the network is unusable
and returns immediate error on each connection attempt
(e.g. ICMP network unreachable error),
but adding backoff for connection failures is out of scope for this change.
In this case connection failure
may be a connection timeout (currently 1 minute),
so it does not make sense to fake idle for another minute immediately after.
However, failure may be immediate if the port is closed
and the server refuses connection every time.
To prevent busy loop in this case
we apply ratelimit to connection attempts rather than login attempts.
This partially reverts ccec26ffa7
If a time value doesn't need to be sent to another host, saved to the db or otherwise used across
program restarts, a monotonically nondecreasing clock (`Instant`) should be used. But as `Instant`
may use `libc::clock_gettime(CLOCK_MONOTONIC)`, e.g. on Android, and does not advance while being in
deep sleep mode, get rid of `Instant` in favor of using `SystemTime`, but add `tools::Time` as an
alias for it with the appropriate comment so that it's clear why `Instant` isn't used in those
places and to protect from unwanted usages of `Instant` in the future. Also this can help to switch
to another clock impl if we find any.
Restart the IO scheduler if needed to make the new config value effective (for `MvboxMove,
OnlyFetchMvbox, SentboxWatch` currently). Also add `set_config_internal()` which doesn't affect
running the IO scheduler. The reason is that `Scheduler::start()` itself calls `set_config()`,
although not for the mentioned keys, but still, and also Rust complains about recursive async calls.
