This adds "stale-while-revalidate" in-memory cache for DNS. Instead of
calling `tokio::net::lookup_host` we use previous result of
`tokio::net::lookup_host` immediately and spawn revalidation task in the
background. This way all lookups after the first successful one return
immediately.
Most of the time results returned by resolvers are the same anyway, but
with this cache we avoid waiting 60 second timeout if DNS request is
lost. Common reason result may be different is round-robin DNS load
balancing and switching from IPv4 to IPv6 network. For round-robin DNS
we don't break load balancing but simply use a different result, and for
IPv6 we anyway likely have a result in persistent cache and can use IPv4
otherwise.
Especially frequent should be the case when you send a message over SMTP
and SMTP connection is stale (older than 60 s), so we open a new one.
With this change new connection will be set up faster as you don't need
to wait for DNS resolution, so message will be sent faster.
If a message from an old contact's setup is received, the outdated Autocrypt header isn't applied,
so the contact verification preserves. But the chat protection breaks because the old message is
sorted to the bottom as it mustn't be sorted over the protection info message (which is `InNoticed`
moreover). Would be nice to preserve the chat protection too e.g. add a "protection broken" message,
then the old message and then a new "protection enabled" message, but let's record the current
behaviour first.
This ensures we do not get stuck trying DNS resolver results
when we have a known to work IP address in the cache
and DNS resolver returns garbage
either because it is a captive portal
or if it maliciously wants to get us stuck
trying a long list of unresponsive IP addresses.
This also limits the number of results we try to 10 overall.
If there are more results, we will retry later
with new resolution results.
Received messages shouldn't mingle with just sent ones and appear somewhere in the middle of the
chat, so we go after the newest non fresh message.
But if a received outgoing message is older than some `InSeen` message, better sort the received
message purely by timestamp (this is an heuristic in order not to break the Gmail-like case
simulated by `verified_chats::test_old_message_4()`). We could place the received message just
before that `InSeen` message, but anyway the user may not notice it.
At least this fixes outgoing messages sorting for shared accounts where messages from other devices
should be sorted the same way as incoming ones.
Otherwise backups exported from the current core and imported in versions < 1.144.0 have QR codes
not working. The breaking change which removed the column is
5a6efdff44.
Before file extensions were also limited to 32 chars, but extra chars in the beginning were just cut
off, e.g. "file.with_lots_of_characters_behind_point_and_double_ending.tar.gz" was considered to
have an extension "d_point_and_double_ending.tar.gz". Better to take only "tar.gz" then.
Also don't include whitespace-containing parts in extensions. File extensions generally don't
contain whitespaces.
There is already code below that emits
progress 0 or 1000 depending on whether
configuration succeeded or failed.
Before this change cancelling resulted
in progress 0 emitted,
immediately followed by progress 1000.
Before this change progress bar only started
when database is already transferred.
Database is usually the largest file
in the whole transfer, so the transfer appears
to be stuck for the sender.
With this change progress bar
starts for backup export
as soon as connection is received
and counts bytes transferred over the connection
using AsyncWrite wrapper.
Similarly for backup import,
AsyncRead wrapper counts the bytes
received and emits progress events.
Instead of treating NULL type error
as absence of the row,
handle NULL values with SQL.
Previously we sometimes
accidentally treated a single column
being NULL as the lack of the whole row.
The "Cannot establish guaranteed end-to-end encryption with ..." info
message can have lots of causes, and it happened twice to us now that it
took us some time to figure out which one it is.
So, include some more detail in the info message by simply adding the
non-translated error message in parantheses.
If we want to put in some more effort for nicer error messages, we
could:
- Introduce one new translated string "Cannot establish guaranteed
end-to-end encryption with …. Cause: %2$s" or similar (and remove the
old stock string)
- And/Or: Introduce new translated strings for all the possible errors
- And/Or: Maybe reword it in order to account better for the case that
the chat already is marked as g-e2ee, or use a different wording
(because if the chat is marked as g-e2ee then it might be nice to notify
the user that something may have gone wrong, but it's still working,
just that maybe the other side doesn't have us verified now)
Info messages are added
at the beginning of unpromoted group chats
("Others will only see this group after you sent a first message."),
may be created by WebXDC etc.
They are not sent outside
and have local Message-ID that
is not known to other recipients
so they should be skipped when constructing
In-Reply-To and References.
Without this fix IMAP loop may get stuck
trying to download non-existing message over and over
like this:
```
src/imap.rs:372: Logging into IMAP server with LOGIN.
src/imap.rs:388: Successfully logged into IMAP server
src/scheduler.rs:361: Failed to download message Msg#3467: Message Msg#3467 does not exist.
src/scheduler.rs:418: Failed fetch_idle: Failed to download messages: Message Msg#3467 does not exist
```
The whole download operation fails
due to attempt to set the state of non-existing message
to "failed". Now download of the message
will "succeed" if the message does not exist
and we don't try to set its state.
HTTPS requests are used to fetch
remote images in HTML emails,
to fetch autoconfig XML,
to POST requests for `DCACCOUNT:` QR codes
to make OAuth 2 API requests
and to connect to HTTPS proxies.
Rustls is more aggressive than OpenSSL
in deprecating cryptographic algorithms
so we cannot use it for IMAP and SMTP
to avoid breaking compatibility,
but for HTTPS requests listed
above this should not result in problems.
As HTTPS requests use only strict TLS checks,
there is no `strict_tls` argument
in `wrap_rustls` function.
Rustls is already used by iroh,
so this change does not introduce new dependencies.
This should fix ad-hoc groups splitting when messages are fetched out of order from different
folders or otherwise reordered, or some messages are missing so that the messages reference chain is
broken, or a member was removed from the thread and readded later, etc. Even if this way two
different threads are merged, it looks acceptable, having many threads with the same name/subject
and members isn't a common use case.
