* put a file without a generated preview in the trashbin
(e.g. a *.docx file)
* open the trashbin
* following errors will show up in the nextcloud.log:
- filesize(): stat failed for ...
- fopen(...): failed to open stream: No such file or directory at ...
- fread() expects parameter 1 to be resource, boolean given at ...
- fclose() expects parameter 1 to be resource, boolean given at ...
- imagecreatefromstring(): Empty string or invalid image at ...
This is because the preview code tries to load an SVG image, which
is obviously only text.
The fix simply handles this before the loading happens and the web UI
keeps showing the default mimetype icon.
OVH has implemented load balancing in a very questionable way where the reverse proxy actually internally adds some cookies which would trigger a security exception. To work around this, this change only checks for the session cookie.
Having two code paths for this is unreliable and can lead to bugs. Also, in some cases Apache isn't setting the PATH_INFO variable when mod_rewrite is used.
Fixes https://github.com/nextcloud/server/issues/983
nextcloud by default uses the `/config/` directory in the source/application tree for its config file(s).
with this commit that directory can be overridden by the `NEXTCLOUD_CONFIG_DIR` environment variable.
in uwsgi, you would use the option `--env "NEXTCLOUD_CONFIG_DIR=/tmp/nx-config/"`
in apache `SetENV …`
and the cli command can be run with: `NEXTCLOUD_CONFIG_DIR=/tmp/nx-config ./occ` (or just use `export` once in the
shell).
NEXTCLOUD_CONFIG_DIR can be supplied with or without the trailing slash (`/`), but in all cases `$configDir` will have
it automatically added if needed.
The other changes are several occurrences of `OC::$SERVERROOT . '/config'` to `OC::$configDir`.
This saves more than 20ms (!) on every request, the previous problem was that `\OC_App::registerAutoloading` calls `\OC\AppFramework\App::buildAppNamespace` which parses the appinfo.xml. Since that was also called multiple times (e.g. on cloud.nextcloud.com over 200 times) that had a significant performance impact. Also on simple PROPFIND requests.
https://blackfire.io/profiles/compare/65a53e6e-7f35-4974-b559-4c81abd01c3b/graph shows the difference nicely.
add section to worklfowengine
enlist only registered sections that also have settings registered to
adjust test
Move admin settings to workflow section
fix wrong var name
Save the container with the app's namespace so we can resolve it
Whenever a group share is created after a direct share, the stime order
needs to be properly considered in the repair routine, considering that
the direct user share is appended to the $subShares array and breaking
its order.
Pick the most recent subshare that has no parenthesis from duplication
which should match whichever name the user picked last. If all
subshares have duplicate parenthesis names, use the least recent group
share's target instead.