This adds our short device ID to the basic auth realm. This has at least
two consequences:
- It is different from what's presented by another device on the same
address (e.g., if I use SSH forwards to different dives on the same
local address), preventing credentials for one from being sent to
another.
- It is different from what we did previously, meaning we avoid cached
credentials from old versions interfering with the new login flow.
I don't *think* there should be things that depend on our precise realm
string, so this shouldn't break any existing setups...
Sneakily this also changes the session cookie and CSRF name, because I
think `id.Short().String()` is nicer than `id.String()[:5]` and the
short ID is two characters longer. That's also not a problem...
lib/fs: Fix conflicts on Android due to fluctuating inode change time
[1] added inode change time to file info in order to support syncing
extended attributes. However, in the case of Android, this inode change
time fluctuates, leading to unexpected conflicts even when the user has
not even touched the files on the Android device itself. Thus, in order
to prevent those conflicts from happening, do not write inode change
time on Android.
[1] 6cac308bcd
Signed-off-by: Tomasz Wilczyński <twilczynski@naver.com>
### Purpose
Deduplicated files are apparently considered 'irregular' under the hood,
this causes them to simply be ignored by Syncthing. This change is more
of a workaround than a proper fix, as the fix should probably happen in
the underlying libraries? - which may take some time. In the meanwhile,
this change should make deduplicated files be treated as regular files
and be indexed and synced as they should.
### Testing
Create some volume where deduplication is turned on (see the relevant
issue for details, including a proper description of how to reproduce
it). Prior to this change, the deduplicated files were simply ignored
(even by the indexer). After this change, the deduplicated files are
being index and synced properly.
This is motivated by the Android app:
https://github.com/syncthing/syncthing-android/pull/1982#issuecomment-1752042554
The planned fix in response to basic auth behaviour changing in #8757
was to add the `Authorization` header when opening the WebView, but it
turns out the function used only applies the header to the initial page
load, not any subsequent script loads or AJAX calls. The
`basicAuthAndSessionMiddleware` checks for no-auth exceptions before
checking the `Authorization` header, so the header has no effect on the
initial page load since the `/` path is a no-auth exception. Thus the
Android app fails to log in when opening the WebView.
This changes the order of checks in `basicAuthAndSessionMiddleware` so
that the `Authorization` header is always checked if present, and a
session cookie is set if it is valid. Only after that does the
middleware fall back to checking for no-auth exceptions.
`api_test.go` has been expanded with additional checks:
- Check that a session cookie is set whenever correct basic auth is
provided.
- Check that a session cookie is not set when basic auth is incorrect.
- Check that a session cookie is not set when authenticating with an API
token (either via `X-Api-Key` or `Authorization: Bearer`).
And an additional test case:
- Check that requests to `/` always succeed, but receive a session
cookie when correct basic auth is provided.
I have manually verified that
- The new assertions fail if the `createSession` call is removed in
`basicAuthAndSessionMiddleware`.
- The new test cases in e6e4df4d7034302b729ada6d91cff6e2b29678da fail
before the change in 0e47d37e738d4c15736c496e01cd949afb372e71 is
applied.
improve parsing of gui-address overrides
make checks for whether the gui-address is overridden consistent by
checking whether the environment variable is set and not an empty
string. the `Network()` function however checked for the inclusion of
a slash instead of the presence of any characters. If the config file's
gui address was set to a unix socket and the gui override to a tcp
address, then the function would have wrongly returned "unix".
the `URL()` function always returned the config file's gui address if a
unix socket was configured, even if an override was specified.
the `URL()` function wrongly formatted unix addresses. the http(s)
protocol was used as the sheme and the path was percent escaped. because
of the previous bug, this could only be triggered if the config file's
gui address was tcp and an unix socket override was given.
simplify the `useTLS()` function's codepath for overrides.
Co-authored-by: digital <didev@dinid.net>
Add a new entry to the unfolded device info to inform the user that the
device has been marked as "untrusted" and all folders shared with it
have to be password-protected or already Receive Encrypted.
Signed-off-by: Tomasz Wilczyński <twilczynski@naver.com>
Add a new entry to the unfolded device info to inform the user that the
device has Auto Accept enabled.
Signed-off-by: Tomasz Wilczyński <twilczynski@naver.com>
Opening and hiding multiple modals at the same time as well as opening a
modal before fully hiding the previous one can lead to the body padding
infinitely increasing by the scrollbar width each time, with the only
way to fix it being refreshing the GUI.
Therefore, always try to ensure to open and hide multiple modals one by
one, and also that the previous modal has fully been hidden before
proceeding to open the next one. The most common case when this problem
happens is when saving config changes which displays a GUI blocking
modal that overlaps, e.g. with folder or device modals that have not
been hidden yet.
Ref: https://github.com/twbs/bootstrap/issues/3902#issuecomment-1547187799
Signed-off-by: Tomasz Wilczyński <twilczynski@naver.com>
Because $scope is missing, there are JavaScript errors when ticking and
unticking the "Untrusted" checkbox in the Advanced tab of the Edit
Device modal.
Signed-off-by: Tomasz Wilczyński <twilczynski@naver.com>
I don't really understand under what circumstances, but sometimes these
calls panic with a "panic: counter cannot decrease in value" because the
value passed to Add() was negative.
Currently, the UI is always blocked from modifications when changes are
being saved, even if the save process takes very little time. This leads
to a situation where showing and closing the blocking modal can take
more time than is actually required to perform the whole operation. The
modal opening and closing very quickly can also cause the screen to
flash for a brief moment, leading to visual discomfort.
Because of this, wait for at least 200 ms and only show the blocking
modal if the changes have not been saved until then yet. The value of
200 ms is loosely based on [1] which states that 'a delay of 0.2–1.0
seconds does mean that users notice the delay and thus feel the computer
is "working" on the command, as opposed to having the command be a
direct effect of the users' actions.' Additionally, the delay must not
be too long, because the main purpose of the blocking modal is to
prevent the user from making further changes, and a longer delay would
possibly allow to do so in that brief amount of time as long as the user
is quick enough with their input.
[1] https://nngroup.com/articles/response-times-3-important-limits
Signed-off-by: Tomasz Wilczyński <twilczynski@naver.com>
In principle a connection can close while it's in progress with
starting, and then it's undefined if we wait for goroutines to exit etc.
With this change, we will wait for start to complete before starting to
stop everything.
gui: Remove unused hard-coded styles from globalChangesModalView modal
Currently, the globalChangesModalView modal has hardcoded th and td
styles. However, they are not even used in the modal itself, because
Bootstrap overrides them with its own styles for these elements in the
same modal. Yet, when hard-coded like that, these styles can conflict
with other table elements in the GUI. Thus, remove them completely.
Signed-off-by: Tomasz Wilczyński <twilczynski@naver.com>
