The PR https://github.com/nextcloud/server/pull/20033 added support
for `fseek` for the S3 storage backend. However, the seek mode SEEK_END
was left out that time. This PR fills this gap.
Signed-off-by: Pauli Järvinen <pauli.jarvinen@gmail.com>
If the first 5 MB are not enough to grab a useful frame for the
thumbnail preview, fall back to reading the full file.
Signed-off-by: Vincent Petry <vincent@nextcloud.com>
In case no distributed memory cache is specified this adds
a database backend for ratelimit purposes.
Signed-off-by: Lukas Reschke <lukas@statuscode.ch>
It's not expected that an app would be unavailable when the app
container is created but when services are registered, but Sentry tells
me on Nextcloud 21 there is an edge case where this can happen.
Therefore this patch hardens the code a bit to log a meaningful error
message and skipping the next code instead of logging a php notice for
the undefined index and an exception for calling a method on null.
Signed-off-by: Christoph Wurst <christoph@winzerhof-wurst.at>
Applies the suggested transformation mentioned in
https://www.php.net/manual/en/migration80.incompatible.php,
> The @ operator will no longer silence fatal errors (E_ERROR,
> E_CORE_ERROR, E_COMPILE_ERROR, E_USER_ERROR, E_RECOVERABLE_ERROR,
> E_PARSE). Error handlers that expect error_reporting to be 0 when
> @ is used, should be adjusted to use a mask check instead
The new code still works on PHP 7, as error_reporting() already
returns 0 when diagnostics are suppressed.
This fixes https://github.com/nextcloud/server/issues/25807 in PHP 8.0.
For PHP 7.x, https://github.com/nextcloud/server/pull/22243 suppresses
the E_NOTICE message from the second session_start() call with the error
suppression operator @, and thus those E_NOTICE messages are still
logged in PHP 8.0.
See also https://github.com/nextcloud/server/issues/25806
Signed-off-by: Chih-Hsuan Yen <yan12125@gmail.com>
The intention obviously was to check whether $lockingCacheClass is defined, and existing class, and available. It was however checked whether the $distributedCacheClass is an existing class, which would have caused an exception already in the previous distributed cache check.
Signed-off-by: MichaIng <micha@dietpi.com>
"parsePhoneNumber()" expects a string, so a TypeError would be thrown if
the phone number value is null.
Signed-off-by: Daniel Calviño Sánchez <danxuliu@gmail.com>
If, for whatever reason, during the loading of a crash reporter a new
log entry is generated, then the lazy loading mechanism will be invoked
*again* while it's already executed. This doesn't result in an endless
recursion, but means that the crash reporters will be built and
registered many times. This then means any further log entry will be
logged x times instead of once.
Unshift makes sure to take the class off the registration list right
away, so another invokation of the same method won't try to do the same
job.
Signed-off-by: Christoph Wurst <christoph@winzerhof-wurst.at>
