Re: Suggestion: port Dalvik JIT to Maemo
 

All this useless rants are the same that we heard about Wine years ago...

Dalvik for Linux (not only Maemo) will be extremely useful when there is no better replacement for a particular application.

The only difference with Wine is that Dalvik is OSS and 'should' be easier to port than reverse-engineering the Windows APIs.

Re: Suggestion: port Dalvik JIT to Maemo
 

Java makes sense in modern C.S. more than C++. Dalvik partly makes more sense than current Java.
Python could conquer, but it needs automatic heuristic static typing JIT VM, and there is none yet.

Qt is in PyQt, QtJambi and why not in QtDalvik.

Meego will fail if it is too dependent on C/C++.

Re: Suggestion: port Dalvik JIT to Maemo
 

If the available dalvik source really is complete, the real difficulty is integrating it with the rest of the system, i.e. wrap the Maemo APIs into Android ones, which can get tricky, esp when it comes to ones that are undocumented and/or closed. I do think it would be slower overall because system processes/launching are done very differently in Android, plus of course the overhead of the wrappers and the occasional headache of replicating NDI. QtDalvik as such to me makes no sense - you don't get the advantages of either Qt or Dalvik, more like just the drawbacks of both.

Re: Suggestion: port Dalvik JIT to Maemo
 

Whats funny is that if the Dalvik port were to happen apps would probably run faster as Android doesn't kill anything and lets them pile up in memory.

Re: Suggestion: port Dalvik JIT to Maemo
 

You assume that all the Apps someone might want to make available could be developed by the Maemo community. For instance, services like Kindle and Audible which provide gateways to paid content on Android and iOS; tell me how you would just make an Audible compatible program? As far as I know its something they have to do, and yes while I have emailed them asking for support, I can't make them do it. So, would it be conceivable that I might want to try to make their Android app run on the N900(yes!)?

Re: Suggestion: port Dalvik JIT to Maemo
 

Yeah, you don't have to wait to pile up, given hildon and Qt, it's already full ! :) But honestly, Android is not too good with that memory management even though it has thrown out all the userspace Linux stuff - having all that Maemo stuff there is not going to help you.

Re: Suggestion: port Dalvik JIT to Maemo
 

WOW wow wow

Re: Suggestion: port Dalvik JIT to Maemo
 

http://www.phonearena.com/news/RIM-t...evices_id16245

Playbook might be using Dalvik. Its such a crazy idea that RIM is even considering it.

Re: Suggestion: port Dalvik JIT to Maemo
 

No is not as crazy as you might thing.

And I have to say I wish Playbook the best on the market. Good native OS and descent API, virtual tools for emulation and development.
Very nice implemented memory management that can handle even multiple (yea Flash) AIR applications, along with native once Java and now this. They will get immediate access to all devs from many platforms and all their experience.

Holy wars between I hate flash or java because I am python or C no longer are relevant. Let's your content to speak for itself and see witch one is going to succeed from end user point of view.

This is my analysis base what I can see around RMI Playbook.
Personally as developer I've never ever considered to buy any BB stuff. But this approach would help me change my mind. And Playbook is on my top list at the moment. I feel tired about those techy wars. what is better or worst. Which OS. Basically there is no the best in all the areas we really wants. As End Users and Developers. because of close parts, companies platform protectionism etc. Once again I see that BB is going to break this wall. Hope so...

You know what?

1. End customer is not giving a s*** about what you thing is better to write a content.
2. Even the best technology in wrong hands can destroy its beauty and you know that even by looking at some "very well ported or written" stuff for N900. Written the best and efficient way possible right? :rolleyes:

Re: Suggestion: port Dalvik JIT to Maemo
 

Even Nokia considers that option - I get an email from project leader - for MeeGo, either the way I'm investigating: compiling dalvik sources against Maemo/Linux API or with a cross-compiler probably from application binaries to Qt.

Re: Suggestion: port Dalvik JIT to Maemo
 

Without going into the speculation what Nokia or Blackberry will do in future products, IMO there is only one reason for adopting Dalvik in any mobile OS - to be able to migrate to Android while keeping the current developer/userbase and their software investment. Why ? Because the moment Android apps/services reliably run on a device and can be bought from the Android Marketplace, there is no reason for any new developer to even think about doing anything native for that particular platform. This is turn means that on the long run you have the same apps/services as your competitors, but you have the monetary/time burden of maintaining a full OS and a compatibility layer IN ADDITION to what every Android vendor does. That's why HTC/Moto/etc do Sense/Blur et al and not take a full OS and put Dalvik on top - it would be an overkill with no benefit.

Re: Suggestion: port Dalvik JIT to Maemo
 

Even after 14 years of using Linux as my primary OS, I keep WINE around for those occasional apps I am forced to use that have no open source equivalent. Presently that is only IE for one particular work related task, all my other needs have been met eventually by a suitable FOSS solution. My point is this is a two way street, and giving users the ability to run necessary software on a foreign system eases migration.

Re: Suggestion: port Dalvik JIT to Maemo
 

WINE is a completely different story. It's (very) far from being a reliable solution for end users, Linux itself does a decent pre-filtering of tech-savvy people, it does not try to integrate with packaging and, last but not least, it is not advertised as a feature for major Linux distros. OS/2 is one example that showed that you *can* get too good in emulating windows. Very few people bothered to write OS/2 specific apps as users could run the DOS or windows version anyway. And then when windows introduced APIs OS/2 couldn't match around Win95, it was game over, as there was no serious native app-base to fall back to.

Re: Suggestion: port Dalvik JIT to Maemo
 

Dalvik is not Android... if you're saying "go dalvik" because you want to run Android applications you're going to be severely disappointed (much in the same way you'd be if you used Preenv to try to run Mojo apps)

And as for the "other" use of Dalvik under other OSs... To be honest I'd be surprised very much if anyone wanted to use Dalvik as a serious development platform. The only reason I could see for that to happen is for legal/licensing reasons, but from a technical PoV Dalvik is far from spectacular...

Re: Suggestion: port Dalvik JIT to Maemo
 

I do understand your concern. I'm just trying to see if you'll consider the benefits of compatibility libraries and emulators as well as the drawbacks. I'll also have to disagree with you on Linux based systems being for tech savvy users only. Not only do we have the N900 itself as proof to the contrary, but I have converted a dozen or so family members and friends to using Ubuntu and none have had any more difficulties using it than they did Windows.

Re: Suggestion: port Dalvik JIT to Maemo
 

Certainly. The bottom line is, however, that providing such a feature that is good for Joe Phone out-of-the box is something that can be done only by the vendor. I understand that (if it worked well) users would value it, developers, too, but as said above, for the vendor, it makes no sense because it's poisoning their (native) ecosystem unless they intend to switch to Android anyway.

It's obviously not a black-and-white picture (yes, I did a fair share of missionary work among my relatives/friends, too), and obviously windows is not *only* used by computer neophytes. However, the average savvy-o-meter is, if I would have to guess, tilted to the Linux side (no, I don't have hard numbers or Stanford studies to prove it).

Re: Suggestion: port Dalvik JIT to Maemo
 

It appears that ymartin59 continues to work away at this. The latest post is in January 2011. It includes a call for help.

Re: Suggestion: port Dalvik JIT to Maemo
 

It looks like someone's going to offer this commercially:

Myriad Alien Dalvik

I guess it is good news, I just wish someone in the FOSS community had done it first.

(EDIT: pointed to the Talk thread, rather than a blog)

Re: Suggestion: port Dalvik JIT to Maemo
 

Hello,
First, I would like to point that the demo only shows usage of display/mouse and network. OK that is a great first step but "porting" all APIs into a "optimized" VM which is not "Dalvik", as any application should be "converted" ?!?
By the way, a commercial product cannot be "incomplete"/"beta", so many tests are required before... so wait.

I get a contact from a Nokia project leader which is interested in the pure Android Dalvik port to standard Linux platform (glibc, X11), a first step before getting runs on Maemo/MeeGo.

Yes, I ask for help because that is a too large work for an hobbist with a full-time job and family.
Please comment on my blog entry http://ymartin59.free.fr/wordpress/i...et-n900-maemo/

Re: Suggestion: port Dalvik JIT to Maemo
 

If the goal = run android apps on maemo/meego/linux, then i see the following paths

1) Port dalvik VM
- Runs some lightweight android apps, but none that require native libraries

2) Port dalvik + android native libraries
- Runs all apps but requires a ton of work to implement / translate android apis to standard linux ones

3) Actual android in a VM / Qemu
- Less work on translation but unusably slow

4) Actual android compiled as a user-mode kernel + chrooted OS

It seems to me that 4 is the best route - does this make any sense?

Re: Suggestion: port Dalvik JIT to Maemo
 

How does Alien Dalvik work? Does it require apps to be converted somehow, or would some Android app package be installed straight to Meego and then just run, or does it need re-packaging or re-compilation?

Re: Suggestion: port Dalvik JIT to Maemo
 

According to what I read, applications must be "converted" - I suppose from binary package but it may be from sources - to run on "optimized" "Dalvik Turbo" implementation of that company.

Re: Suggestion: port Dalvik JIT to Maemo
 

I am in Same problem , I am trying to run Android apps on Meego.
As I just started I want to know which Approach will be better
Porting DVM + android Native libs - my point of view this is good solution in long run but definitely require huge work at each level of android as well meego side.

Android as usr-mode kernel + Chrooted OS - good starting point to arrive some conclusion. but android expects multiple user while Linux expect sing user. Every android app installed has new UID handling this will be tricky.
Canaonical trying to do this -https://wiki.ubuntu.com/Specs/AndroidExecutionEnvironment


I will keep updated. any suggestions help are always welcome.