I'm working on creating a CordApp which I'm expecting to run on user's phone. Android as a starting point. From my reading so far, Android phones do not have a JVM running on the phone and the compiled code is converted into either Dalvik or something similar.
Has someone tried to install their CorDapp on an android based mobile device?
I'm a nooby in the area of CordApp as well in android apps but hoping that I can find some good starting tips here. Thanks for help.
The bottom line is that Corda needs a JVM environment to execute. The most light-weight test we have tried is Raspberry Pi with 2G Memory.
(answering the follow-up in the comment above)
You might be able to run several distinct nodes on some server. However at that point there would be little point in using Corda in the first place.
Once you run a node "somewhere else" the user won't own the private key for its node, and all the safety of using Corda disappears. You're reverting to a 'trusted third party' model.
Lately, while attempting to build an interesting open source android-web streamer: Android-Eye, I noticed that several people, including myself have been stumped by process of building the application given that it contains c/c++ code that (I think) needs to be compiled by the android ndk before anything can work. The fact that the dev doesn't seem to be active anymore doesn't make this any easier.
I've been trying to get this to work for well over 3 days now and can say with full confidence that I do not posses the knowledge (especially for the ndk) to get this thing built. The long list of things i've tried is too extensive to be covered in this post but here are a few git issues posted by others that have experienced similar issues as me:
https://github.com/Teaonly/android-eye/issues/32
https://github.com/Teaonly/android-eye/issues/27
I understand that it may look as if I am asking to be "spoonfed" the process of building the application but a few pushes in the right direction would be excellent :)
BTW, I have access to windows and mac boxes incase that influences the build process.
Thanks
I need to do some menial batch tasks on my phone, and I don't want to jump through all the hoops of making an "app" with a GUI and all that just to do them (the tasks are of the type you'd hack together in BASH in five minutes on a sane system). I can't seem to find any place on the net that explains how to simply make an ordinary program (in any language, but Java is OK if that eases interaction with Android) with access to the Android API that can simply be run by SSHing into the phone and running it as a normal process. No need for an APK package, no need for a GUI, no nothing. This should be the simplest thing in the world, but every example out there seems to be first and foremost concerned with making a GUI and working with Eclipse and the SDK instead of doing the basics first.
Any tips?
(I know this is probably borderline SuperUser, but then again, there's a programming question at the bottom: How do you make an ordinary (Java) program that can be run from the terminal on an Android phone and still use the API?)
Here : Running a shell script on android device using adb
and : https://play.google.com/store/apps/details?id=stericson.busybox&hl=en
and : http://forum.xda-developers.com/showthread.php?t=537827
and : http://strawp.net/archive/recipe-for-a-decent-bash-shell-in-android/
and : http://digitaldumptruck.jotabout.com/?p=938
I can't seem to find any place on the net that explains how to simply make an ordinary program (in any language, but Java is OK if that eases interaction with Android) with access to the Android API that can simply be run by SSHing into the phone and running it as a normal process.
That's because it's not especially important to the bulk of Android users or developers.
How do you make an ordinary (Java) program that can be run from the terminal on an Android phone and still use the API?
You are welcome to use the Scripting Layer for Android to write some scripts, but you have limited access to the Android SDK, and they cannot "simply be run by SSHing into the phone". This is supported by the SL4A team.
You are welcome to experiment with the dalvikvm command, though off the top of my head I do not recall whether or not it is available on production devices, and I do not know if it can "simply be run by SSHing into the phone". And, bear in mind that using this is completely unsupported.
You are welcome to write your own C/C++ code for ARM (or whatever CPU architecture your device runs). This "simply be run by SSHing into the phone" but has no access to the Android SDK.
I still cannot believe that that kind of stuff isn't on the first page of every Android development introduction out there.
There are over 200 million users of Android devices. What percentage of those users do you think want to
"make an ordinary program... with access to the Android API that can simply be run by SSHing into the phone and running it as a normal process"? 0.01%? 0.001%? My money is on 0.0001%.
The "first page of every Android development introduction out there" should be focused on stuff that matters to closer to 100% of the user base. You, of course, are welcome to build up your own site focused on this sort of thing, to cater to those users who are interested in creating these sorts of programs.
From http://git-annex.branchable.com/design/assistant/blog/day_184__just_wanna_run_something/:
While I already have Android "hello world" executables to try, I have
not yet been able to run them. Can't seem to find a directory I can
write to on the Asus Transformer, with a filesystem that supports the
+x bit. Do you really have to root Android just to run simple binaries? I'm crying inside.
It seems that the blessed Android NDK way would involve making a Java
app, that pulls in a shared library that contains the native code. For
haskell, the library will need to contain a C shim that, probably,
calls an entry point to the Haskell runtime system. Once running, it
can use the FFI to communicate back to the Java side, probably. The
good news is that CJ van den Berg, who already saved my bacon once by
developing ghc-android, tells me he's hard at work on that very thing.
and some specific advices in the comments below:
See http://kevinboone.net/android_nonroot.html for info on where in
the android filesystem you have write, exec ability.
Basically you have these abilities in /data/local from adb shell (and
in debuggable app's folders using run-as with adb shell), and in
/data/data// for each app (for example the terminal emulator's
data dir when using the terminal emulator).
...
http://git-annex.branchable.com/design/assistant/blog/day_185__android_liftoff/:
Thanks to hhm, who pointed me at KBOX, I have verified that I can
build haskell programs that work on Android.
http://kevinboone.net/kbox.html:
KBOX [...] gives you the terminal emulator, a
decent set of Linux utilities (supplied by busybox), ssh and rsync
clients and servers, and a few other things. In addition, there are a
number of add-on packages for expanded functionality.
Well, it's just about running an executable on Android, and not about writing an executable that would access Android API...
I can't seem to find any place on the net that explains how to simply make an ordinary program [...] with access to the Android API that can simply be run by SSHing into the phone and running it as a normal process.
An answer, translated from a note by vitus-wagner:
Termux is an advanced terminal emulator plus lots of Unix-like software with command-line interface (in packages managed by APT). Actually, not only CLI (command-line), but also GUI as well (though the GUI software not tried yet).
Unlike the way of the various popular "linux deploy" (which make something like a container, at least a chroot, with things installed into directories according to the traditional filesystem hierarchy), Termux seems to aim at integrating into the host system. For this purpose, it has a plugin, Termux:api which is able to do a lot of interaction with the system: open a file in a native Android app, send an sms, take a picture with the camera, or even say something by means of the system TTS engine.
There are many more addons -- see wiki.
(A side note. An integration like that could be expected--if not from MSYS--from GnuWin32, but there is nothing close to Termux under Windows w.r.t. the degree of integration.
However, for some strange reason, people are asking much more about how to make it more "Linux-like" on the forum, rather than how to use it effectively to solve smartphone-specific tasks...)
A toolkit for cross-compilation is available, so that one can try to package his favorite software.
Actually, it is able to do compilation locally on the device, but it seems not to be able to make a package locally.
Some things to know:
One needs Hacker's keyboard or something similar. One can't live here without Esc, Tab, Control. Or one could try to learn the Touch Keyboard.
vim ran with an encoding different from utf-8, and the Russian letters were displayed incorrectly. So, set encoding=utf-8 had to be written in .vimrc.
ssh to another computer at home couldn't login. The reason was simple: it used the username u0_a95 instead of one's usual username. (One can write User your_username in .ssh/config to permanently "fix" it.)
I'd recommend doing a research on XDA-Developers board
Hy everyone, I'm Korean and a little short on using english so please try to understand if I say things not appropriate.
So, my status is that I have odroid-s.
What I'm trying to do for like month or more is that I want to make android to HelloWorld.
What I want to say is that, on the odroid-s, bootloader part, kernel part is the same but the framework part(which will be android), I'll remove all the android part and replace it with just HelloWorld program. The purpose of this HelloWorld program is to display HelloWorld on the screen.
What I think I discovered is that, as I 'vimdiff' bootlogs between normal bootlog and the one that I removed all the system partition part (which is android system partition part) is that android kernel's init goes on and executes console(/bin/sh), netd(bin/netd), ... and it enables adb and it completes his work.
So my conclusion is, I need to use Linux kernel that is non-modified and modify it for odroid-s, and use that kernel for my HelloWorld program!
What I want to ask is.... Am I doing it right?? T_T
My goal right now is to make HelloWorld come out from this odroid-s device...
Please somebody help me. If anybody don't understand what I wrote plz tell me, I'll fix it.
Thx for reading....
The modifications to the linux kernel are likely to be irrelevant to your goals, so you might as well leave them in place for simplicity.
Your biggest challenge is going to be output - where do you want to send it? If you had one of the devices that has (or can have with the right kernel config) a debug serial port, then it would be really easy to write something triggered by the init script (or even use 'echo' in the script) which outputs your message on that port.
But chances are you want to put something on the screen. This is going to be overwhelmingly more complicated, and perhaps device dependent. The way the android runtime does this for actual apps is going to be way more involved than you probably want to get into.
A more practical approach might be to look at how the boot animation is done. For starters you could just replace it with a static image that says "hello world". Once you can do that, the next step would probably be to find some character generator code. Finally you might want to implement scrolling and other terminal-like features.
As an alternative approach, there are builds of more traditional linuxes for some android devices - debian or ubuntu for example. These may include console implementations capable of displaying on the device screen.
As another idea, if you are flexible about how much of android you would be willing to leave on the device, you could build a version of the android terminal emulator example, modified to be a home screen replacement. You might be able to remove a lot of android components (eventually including the default home screen). Or on a secured device (ie, most consumer devices that haven't been rooted) you could just do the home screen replacement while leaving the actual system unmodified. It wouldn't be secure against users wanting to run other things, but generally the user would interact only with your code.
I need to profile dynamic library calls of android application and have decided to profile using "ltrace" tool for this purpose. I combined "ltrace" tool into android package, then compiled together successfully. It works fine with shell program just like usual linux console application.
However, I cannot properly profile android applications, which are forked from "Zygote" process. I tried attaching ltrace to "Zygote" process to follow child processes of it, but I only got SIGCHLD and SIGSTOP signals and terminated abnormally, showing the following errors.
unexpected instruction 0xffffffff at 0xffff0508
I'm just wondering if anyone has ever tried this kind of profiling on android system. Any short comments can be very helpful for me.
Thank you in advance.
Try attaching ltrace to your android application's process after it is it forked off zygote. You will need an ltrace built for android's bionic libc, and one that has reasonable handling of threads.
There is a way to set a debuggable android app to wait for connection of the java debugger, you could use that, connect ltrace, and then connect and disconnect the java debugger to start it going again. This should capture most of your own logic, though not the entirety of the startup.
IIRC you are under a time limit to connect and get it going again, otherwise it may resume on its own to avoid triggering an application not responding situation.
It may be that you will learn more by using the java debugger to see what is going on, and then reading the source to see how that is implemented on the native side.
Mr. Chris Stratton's way will work.
LoadLibrary() is another point to break for debbuging jni.