Remove scanAll which relies on the "shareinfo" endpoint that returns the
full cache tree.
The latter can become big for big shares and result in timeouts.
Furthermode, the full tree would be retrieved again for each and every
detected change which can become expensive quickly.
Signed-off-by: Vincent Petry <vincent@nextcloud.com>
In https://github.com/nextcloud/server/pull/28774 we disabled the
caching for the groupfolder application since it worked due to the fact
that in groupfolders, getFileIds could be called with the same $cacheId
and path for actually different groupfolders.
This revert this change and instead add the folderId from the
groupFolder to the cacheId. This solve the issue of the uniqueness of
the cacheId inside GroupFolder. Downside is that we introduce
groupfolder specific implementation inside the server repo.
The seconf optimization is to not consider paths starting with
__groupfolders in executeCheck. This is due to the fact that files in
the groupfolder application call two times executeCheck one time with
the url __groupfolder/<folderId>/<path> and the other time with <path>.
The first time will always return an empty systemTags array while the
second call will return the correct system tags.
Signed-off-by: Carl Schwan <carl@carlschwan.eu>
This will only respect the setting inside the file app. For other apps
we will either need to do an API call or add an input field with the
same idea to spare a blocking api call.
Signed-off-by: Carl Schwan <carl@carlschwan.eu>
(cherry picked from commit 1fa58be1aa)
Use a backup table to copy the data, drop table and recreate it with
correct primary key, then copy the data back and drop the backup table.
Signed-off-by: Côme Chilliet <come.chilliet@nextcloud.com>
for the following query 'SELECT "path" FROM "oc_filecache" WHERE ("storage" = $storage) AND ("size" < 0) ORDER BY "fileid" DESC LIMIT 1;'
currently the database will in some cases decide to priorize the sort by fileid over the filter when picking what index to use, resulting in a much slower query.
by creating an index that allows first sorting by fileid and also filter by storage and size this case will be greatly sped up
Signed-off-by: Robin Appelman <robin@icewind.nl>
Whenever an error occurs, also hide the progress bar.
The logic was also adjusted to properly detect uploads that are pending
deletion, in which case the progress bar can already be hidden.
Signed-off-by: Vincent Petry <vincent@nextcloud.com>
Add a new approach for flagging an upload as aborted because we can't
rely on the browser fully cancelling the request as we now seem to
receive an error response from the server instead of a jQuery "abort"
message.
Signed-off-by: Vincent Petry <vincent@nextcloud.com>
When using the browser back button or clicking on sections on the left
sidebar (like favorites), the "changeDirectory" jQuery event did not get
called, so apps like recommendations would not notice the directory
change.
This fixes the issue by also setting changeUrl to true when the file
list's directory got changed as a result from a URL change.
Added optional changedThroughUrl argument to make sure the event
recipient knows if the change was done through a URL change and make it
possible prevent a loop in the onDirectoryChange handler that actually
changes the URL when the origin was already from a URL change.
Signed-off-by: Vincent Petry <vincent@nextcloud.com>
For most image formats, the header specifies the width/height.
PHP allocates an image object from that size, even if the actual
image data is much smaller. This image object size is not limited
by the limit configured in PHP.
The memory limit can be configured through "config.php" setting
"preview_max_memory" and defaults to 128 MBytes which should be
enough for most images without filling up all memory.
Signed-off-by: Joachim Bauch <bauch@struktur.de>
John Molakvoæ
Nextcloud bot
Anderson Luiz Alves
Vincent Petry
Carl Schwan
Arthur Schiwon
Joas Schilling
szaimen
Côme Chilliet
John Molakvoæ
MichaIng
Louis Chemineau
Robin Appelman
Julius Härtl
Joas Schilling
Joachim Bauch