Josh Richards [Wed, 26 Jun 2024 13:31:44 +0000 (09:31 -0400)]
fix(config): Add missing handling for envCache in getKeys()
NC_ env variable overrides were not appearing in
the output of `occ config:list system` nor `occ
config:system:get xxx`. This was creating nearly
impossible to diagnose configuration/ behavior
disprepancies.
- Refactored readData() so that we aren't saving
the entire environment in the envCache anymore
(only those prefixed "NC_") and so that we save
NC_ provided config values under their real
key.
- Refactored getValue() to accommodate readData()
refactor
- Fixed getKeys() to properly return
envCache keys too
Environment provided config variables now appear
in `occ config:list system` as expected.
Environment provided config variables now appear
when queried via `occ config:system:get KEY`
Louis Chemineau [Mon, 16 Sep 2024 09:10:31 +0000 (11:10 +0200)]
feat: Reset route if neither the Viewer of the Sidebar is open
When the viewer or the sidebar is opened, we add the fileid to the route.
When both of them are closed, we do not remove the fileid from the route.
This means that, upon reload, the sidebar will be opened even though it was closed previously.
This PR ensure that the fileid is removed from the route when both the Sidebar and the Viewer are closed.
fix: Do not try to set HTTP response code on already closed connection
This avoids a PHP warning in the logs about trying to set the response
code while the output already started. It’s useless to try to print an
error page anyway in this situation because the connection was closed
already.
fix(config): Throw PreconditionException always when it didn't match
Previously even when the precondition did not match, the call "passed"
when the after value was the expected one. This however can lead to
race conditions, duplicate code excutions and other things.
Ensure that after a right-click the context menu position is resetted,
so that pressing the "actions"-menu button it is displayed on the correct location.
Signed-off-by: Ferdinand Thiessen <opensource@fthiessen.de>
Côme Chilliet [Tue, 27 Aug 2024 10:53:38 +0000 (12:53 +0200)]
feat(transfer-ownership): Correctly react to encrypted files
For E2EE encrypted files, we abort the transfer.
For SSE encrypted files, we abort only if not using master key.
Also fixed the check for when the path to a single file is used.
fix: Hide status tooltip in row to add a new mount point
The row to add a new mount point is cloned when a new mountpoint is
added, so it is expected that it includes a status span. However, it
should not be displayed in that row, only in the cloned row when its
status is updated.
Signed-off-by: Daniel Calviño Sánchez <danxuliu@gmail.com>
fix: Fix unmodified placeholder replacing the actual value when updating
When updating global storages and user storages a property is not
updated by "StoragesService::updateStorage()" if the value matches the
unmodified placeholder. However, userglobal storages are not updated
through the "StoragesService"; as only the authentication mechanism is
updated it is directly done with "saveBackendOptions()" in
"IUserProvided" or "UserGlobalAuth". Due to this the unmodified
placeholder value needs to be explicitly checked in those cases and
replaced by the actual value (note that in this case it is not possible
to just skip updating a specific property).
Signed-off-by: Daniel Calviño Sánchez <danxuliu@gmail.com>
test: Add integration tests for saving external userglobal storages
As the external storage uses the Nextcloud server itself the number of
workers of the PHP process running the Nextcloud server had to be
increased. Otherwise if a request is sent for the external storage while
handling a request from the integration tests a deadlock would occur.
Signed-off-by: Daniel Calviño Sánchez <danxuliu@gmail.com>
fix: Reset selected backend when adding a new storage
As a new storage is added by selecting a backend the selected backend
needs to be reset. Otherwise it is not possible to add another storage
with the same backend.
Signed-off-by: Daniel Calviño Sánchez <danxuliu@gmail.com>
fix: Restore default status tooltip when no status message is provided
If the status is updated but no explicit message is provided (for
example, if the status check succeeded) the default tooltip (from the
template) is now set to prevent a mismatch between the status and the
tooltip (for example, if the configuration is fixed after a failed
status check).
Signed-off-by: Daniel Calviño Sánchez <danxuliu@gmail.com>
fix: Set status tooltip to error message on failed actions
When saving, updating and rechecking an storage fails (which is
different to the soft-fail when the action itself succeeds but the
status check does not) further details are provided in the error message
of the response, which is now set as the tooltip.
Signed-off-by: Daniel Calviño Sánchez <danxuliu@gmail.com>
fix: Set status tooltip to status message when saving an storage
When a storage is saved the status check can fail even if saving the
storage succeeds. In those cases further details are provided in the
status message of the storage, which is now set as the tooltip,
similarly to how it is done when rechecking the storage.
Signed-off-by: Daniel Calviño Sánchez <danxuliu@gmail.com>
fix: Remove status check when configuration was changed
Setting a null status was supposed to remove the status check, but
nothing was changed in that case. Now the status check is properly
removed, and doing that by hiding the element rather than just turning
it invisible also prevents that clicking on the invisible status
triggers a check, as until the new configuration is saved the check will
still be performed with the old configuration, which could be misleading
for the user.
Additionally, an explicit width is set to the parent of the span element
to prevent its width from changing when the span is shown and hidden.
Signed-off-by: Daniel Calviño Sánchez <danxuliu@gmail.com>
In Nextcloud 29 and later the tooltip usage threw an error that caused
the UI to be unusable, so it was removed. In Nextcloud 28 there is no
error, but using the jQuery tooltip call causes the tooltip to be
unreadable, so it was removed as well (which makes the native tooltip to
be shown instead).
Signed-off-by: Daniel Calviño Sánchez <danxuliu@gmail.com>