Open GoogleCodeExporter opened 8 years ago
A couple of other ideas occurred to me after I wrote the original request:
4) Checking several binaries during application startup from the UI thread may
cause "red flashes" due to blocking disk I/O. Should RootTools provide a
non-blocking option which starts a background thread, and optionally issues a
callback on the UI thread when finished?
(Is it safe to access raw resources from another thread?)
Or maybe RootTools should just trust the caller to run this operation from an
AsyncTask.
5) In many cases, apps will want to extract several resources at the same time.
Is it worth taking a list of (resid, filename, mode) tuples and handling them
all at once, returning a code indicating:
- Success, no updates
- Success, some/all updates
- Failure on one or more resources
6) Many apps bundle separate sets of binaries for <=ARMv6, ARMv7, x86, and
MIPS. So if #5 is implemented, maybe the tuple looks more like:
(arm_resid, armv7_resid, x86_resid, mips_resid, filename, mode)
with RootTools checking the relevant system properties to figure out which one
to copy over.
Original comment by cerne...@gmail.com
on 10 Sep 2013 at 5:32
I have been meaning to introduce something like this for quite a while.
You actually push it beyond what I had thought about. I know that we will
implement #1 and #2 at the very minimum.
-Chris.
Original comment by militate
on 16 Sep 2013 at 8:29
One more request:
7) Installer.installBinary() currently invokes Shell.startRootShell() to change
perms on the extracted binary. If we're just extracting to
/data/data/app.package/files/, it should not be necessary to "su". Changing
this behavior would make installBinary() usable by non-root apps.
Original comment by cerne...@gmail.com
on 23 Sep 2013 at 5:09
A small step in the right direction: installBinary() now updates a binary if it
finds its MD5 checksum has changed.
Original comment by militate
on 27 Sep 2013 at 2:34
Original issue reported on code.google.com by
cerne...@gmail.com
on 8 Sep 2013 at 6:56