Drop the user-visible separation, make everything just Tuhi. Directory
structure remains largely separate (tuhi vs tuhigui) except for the po files
which are in the top-level directory because convention.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
This is a plain subtree merge without any changes to the tuhigui tree itself.
As we'll keep the distinction between the DBus server and the GTK UI, the
directory layout will reflect that.
No-one but Arch packages Tuhi at the moment, so let's assume it's not
running/installed, install it and start it up together with TuhiGui and voila,
everything just works as it should be.
A user may want to download the drawing multiple times. And besides, after
switching back to the default icon theme, I couldn't find a decent icon
anyway.
Enable the delete button that's been hidden away so far. When clicked, we
suffix the cached json file with '.deleted'. This also causes a popup with an
undo button to appear (taken from Nautilus). When clicked, that button will
restore the drawing again.
This is really all just renaming anyway, because nothing ever gets deleted
here.
An extra filter is needed to skip Tuhi drawings that have been deleted in the
GUI - we don't want to add the drawings we just deleted again just because
Tuhi still keeps them in cache.
Tuhi caches the json data but has no guarantee about storage. So we cache the
json we get from Tuhi and store it in our settings dir through our Config
backend. Then we use those cached values to generate the SVG files.
This is saved for posterity in the new settings file, so we need a new Config
object (singleton is enough here).
For now, whenever the orientation changes, we just wipe our UI and re-generate
all SVG files. Much easier than messing with proper SVG rotation given that a
device should only ever change between rotations once.
Previously we only ever updated on the initial device assignments. Especially
when the device is offline while started, this means we never get the right
icon.
This broke device initialization in some cases (not sure why it worked until
today though). Where a device is in register mode but the ManufacturerData has
not yet updated to the 4-byte string (i.e. a device previously known
normally), the device would now always call listen(). The firmware - in
register mode - would start initializing the connection though and get
rejected, leaving us with a device that cannot be registered.
This reverts commit a061240b11.
Peter Hutterer