It now supports only one storage of each type. Only one Storage could be
loaded to the memory as well.
Options' Cloud tab now changes the Storage only when user pressed OK
button, giving the ability to look through the Storages without actually
changing them.
GoogleDriveDownloadRequest, which resolves file id and then downloads it
with GoogleDriveStorage::downloadById().
GoogleDriveStreamFileRequest, which resolves file id and then returns
file stream with GoogleDriveStorage::streamFileById().
This commit also adds GoogleDriveStorage::streamFileById() itself.
A minor GoogleDriveResolveIdRequest fix added.
With these one can download files from Google Drive.
Because of the Google Drive StorageFile now contains yet another field,
`id`. For other storages `id` == `path`, and thus all common Requests
(such as SavesSyncRequest, DownloadFolderRequest, etc) must be using
id() instead of path(). That way these Requests won't cause id resolving
which could be quite slow (when you call it for all files in the folder,
for example).
Now openForLoading() and openForSaving() check whether file is locked,
so AGOS and SCUMM engines Ctrl+number and Alt+number hot keys shouldn't
be able to save/load in these slots during saves sync.
Engines with "simple" savenames would support "Run in background" in
save/load dialog and gradual save slots unlocking. Other engines
save/load feature would be locked until save sync is over.
Now it also creates the "base" ScummVM directory if there is no such
directory yet. This way SavesSyncRequest works fine when no "ScummVM" or
"ScummVM/Saves" folder exist in the Google Drive.
It now keeps newly added Requests in separate array, so iterators don't
break when one adds a Request while ConnMan iterates its array.
The array is also shielded with mutex.
When listing directories, you get a list of StorageFiles, which path()
is actually Google Drive id. Thus, if you list a directory recursively,
you won't be able to determine whether all files are within one
directory or have some hierarchy. I'd fix that as soon as it would be
needed.
GoogleDriveResolveIdRequest gets a lowercase path and searches for the
specified file's id. To do that it lists path's subdirectories one by
one with GoogleDriveListDirectoryByIdRequest.
GoogleDriveListDirectoryByIdRequest gets a Google Drive id and lists
that directory (not recursively).
MetaEngines don't get "locked" files in the list, so won't try to open
these.
Save/Load dialog updates save list every time SavesSyncRequest tells it
to.
It's shown by SaveLoadChooserDialog when files are downloaded and some
save slots are locked. One can hide that dialog to interact with
non-locked slots or cancel saves sync completely. Dialog's label shows
current sync progress.
Dialog automatically hides itself when all files are downloaded.
WARNING: right now that results in a crash!
CloudIcon is now a Request which is automatically added once first
Request is added to ConnMan. When icon decides it should disappear, it
gets FINISHED, so ConnMan would switch off the timer if it was the last
Request.
I didn't like how FINISHED Requests were waiting until the next
interateRequests() call to be removed when we could easily remove those
after they changed their state in their handle().
To achieve smoother animation, ConnectionManager's timer now is 20 times
more frequent.
I'm encountering some strange libcurl.dll segfault problem when I close
the application while some Requests are active. It's not
CloudIcon-related, so it's more likely related to this 20 FPS timer.
This problem shows up only in Visual Studio for me.
Alexander Tkachev