[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
| 
Re: [dsdp-mtj-dev] Preverification support on MTJ
 | 
Gustavo,
I think we probably need to have a conversation with the larger group 
concerning the scope of toolkit support that MTJ should support out of 
the box.  I can see concerns with having an all-inclusive set of 
toolkits.  With that said, I'd like to see us support at least one of 
the all-Java toolkits out of the box.  It appears that MPowerPlayer has 
stagnated and that microemu is probably the better option at this 
point.  To support one of these toolkits is going to require some kind 
of support for preverification on those platforms.  The advantage of an 
all-Java solution (emulator and preverification) means that we can 
support pretty much any development platform.
The OSX preverifier can be downloaded as part of the MPowerPlayer 
toolkit.  In addition, I managed to compile the preverifier from the 
phoneme project without too much trouble.  I think there are some decent 
options there...
Craig
Paula Gustavo-WGP010 wrote:
Hi craig,
Some comments below
:)
gep
-----Original Message-----
From: dsdp-mtj-dev-bounces@xxxxxxxxxxx [mailto:dsdp-mtj-dev-bounces@xxxxxxxxxxx] On Behalf Of Craig Setera
Sent: sábado, 17 de maio de 2008 17:26
To: Mobile Tools for The Java Platform mailing list
Subject: Re: [dsdp-mtj-dev] Preverification support on MTJ
If we don't support non-UEI SDK's, there are two significant (in my
mind) problems:
1) We walk away from Mac because it does not currently have a UEI emulator.
2) We remove functionality that EclipseME had.
I think both of these are bad choices.  #1 because there is a growing Mac user base (including myself) and #2 because I'd like to transition to MTJ exclusively and yet I feel it is necessary to have the full EclipseME functionality in MTJ for me to do that.
[gep]: I agree that we want to support mac, but I think that we need to consider the boundary between the SDKs and MTJ. Let's say that we want to support solaris and there is no MIDP SDK on solaris at all, does MTJ need to provide a MIDP/CLDC implementation on solaris? I wouldn't like to do that :)). That's why I'm not totally comfortable to maintain on MTJ a code that is "suppose" to be provided by the SDK (but I can change my mine on that :))
In the long term, I'd like to see the embedded all-Java preverifier get fixed up, tested and working.  I'm interested in doing that work, although I have no idea when I will find the time.
[gep]: if you are interested in doing that work then I'm ok with that.
In the mean time, I think it would be OK to require the MPowerPlayer osx preverifier for Mac users.  In doing this, it would be necessary to allow the specification of an arbitrary path to the preverifier rather than assuming that it is tied to the location of the emulator.  This would require some changes in the core as well as in the UI.
[gep]: I haven't saw this osx preverifier yet, but, if there is already one there we can use, we can add the support to it. If you have some information about it, can you send to us? In the meantime I took some time to look at current MTJ preverification code and also at proguard preverifier. I did an initial prototype that use proguard preverifier and it seems to work ok. I attached a patch on https://bugs.eclipse.org/bugs/show_bug.cgi?id=234139 with the code. I didn't committed to svn, since there is no agreement yet that this is a good feature (and it is still a prototype). We can continue the discussion about this on the bugzilla.
Those are my thoughts.
Thanks,
Craig
PS - After talking to Eric LaFortune about the proguard preverifier, he thought it might be heavyweight to use as a pure preverifier (although possible).  He seemed to imply that it was more meant to be used in a scenario where you want to obfuscate and preverify at the same time.
On May 15, 2008, at 12:34 PM, Paula Gustavo-WGP010 wrote:
  
Hi guys,
I'm back from javaone and I'm trying to catch up with all the 
discussions on mtj list. There are a lot of emails :)
About this preverifier question. As far as I know, the preverification 
that was added to java 1.6 is different from the one that is used on 
CLDC, since that we can not just use java 1.6 compiler to preverifier 
the CLDC classes 
(http://forums.java.net/jive/thread.jspa?messageID=192443
).
I think that this is more a requirements issue. Do we want to have 
that on MTJ or not??? I understand that we want to support different 
platforms (linux, mac and windows), but based on craig's meail I think 
that the real question is different:
Do we want for now to support non-UEI SDKs???
If we have a decision to support only UEI compliant SDKs then each SDK 
is suppose to provide their preverifiers on each platform (linux, mac 
or windows) and we don't need to add a preverifier on MTJ.
The reason that EclipseME added this features was the supported to 
those 2 SDKs that are not UEI compliant and do not include a 
preverifier (please corret me if I'm wrong craig). Since that 
eclipseME added its own to fill the gap of those SDKs.
One possible plan to address the non-UEI SDKs is to use proguard 4.x 
that includes a java preverifier (I never tested it, but it probably 
works well). We can:
- document that in MTJ help
- add some UI that the user can configure proguard to preverify the 
code
- remove our own preverifier implementation
I'm not sure if a like the idea to have our own preverifier on MTJ 
(and also to maintain it), since most SDKs include their own and also 
it is possible to use other solutions like proguard that can be 
downloaded and configured on our UI (but I'm open to change my opinion 
:) ).
Comments???
:)
gep
-----Original Message-----
From: dsdp-mtj-dev-bounces@xxxxxxxxxxx 
[mailto:dsdp-mtj-dev-bounces@xxxxxxxxxxx
] On Behalf Of Hildum Eric-XFQ473
Sent: terça-feira, 13 de maio de 2008 23:40
To: Mobile Tools for The Java Platform mailing list
Subject: RE: [dsdp-mtj-dev] Preverification support on MTJ
If I understood the context correctly, the point of the enhancements 
to the Java 6 compiler was that it would preverify the code for J2ME 
and Java 1.4 targets (which presumably should work for all J2ME 
implementations...).
Eric Hildum
Senior Product Manager, Mobile Developer Tools & SDK Software 
Platforms and Delivery Ecosystem and Market Development Motorola
Direct: +1-408-541-6809
Mobile: +1-510-305-0801
809 11th Avenue
Sunnyvale, CA 94089
USA
-----Original Message-----
From: dsdp-mtj-dev-bounces@xxxxxxxxxxx 
[mailto:dsdp-mtj-dev-bounces@xxxxxxxxxxx] On Behalf Of Craig Setera
Sent: Tuesday, May 13, 2008 16:28
To: Mobile Tools for The Java Platform mailing list
Subject: Re: [dsdp-mtj-dev] Preverification support on MTJ
My understanding is that while the concepts are the same between ME 
and SE, that the actual code attributes that are generated are 
different.
It may be close and we may be able to "fudge", but for now, I don't 
think it is going to work for us.  The other piece of that puzzle is 
that a lot of the devices don't like code that is compiled beyond 
version 1.2 of Java... Thus, switching to Java 6 isn't going to fly in 
that regard.
Craig
On May 12, 2008, at 10:56 PM, Hildum Eric-XFQ473 wrote:
    
I have recently heard that the Java 6 compiler contains the 
capability
      
to build preverfied code directly through compile time switches. I am 
not sure if this capability may help us. If this capability is also 
built into the Eclipse Java compiler, then we may be able to sidestep 
the preverifier completely. That would clear the way for Mac and 
Linux
      
support in MTJ.
Eric Hildum
Senior Product Manager, Mobile Developer Tools & SDK Software 
Platforms and Delivery Ecosystem and Market Development Motorola
Direct: +1-408-541-6809
Mobile: +1-510-305-0801
809 11th Avenue
Sunnyvale, CA 94089
USA
-----Original Message-----
From: dsdp-mtj-dev-bounces@xxxxxxxxxxx 
[mailto:dsdp-mtj-dev-bounces@xxxxxxxxxxx] On Behalf Of Craig Setera
Sent: Sunday, May 11, 2008 19:13
To: Mobile Tools for The Java Platform mailing list
Subject: Re: [dsdp-mtj-dev] Preverification support on MTJ
Eric,
Sorry for not getting a response back on this for a while.  Part of 
it
      
is that I've been busy and part was that I was doing a bit of 
experimenting.
On Apr 28, 2008, at 11:30 AM, Hildum Eric-XFQ473 wrote:
      
I have two questions for multiplatform support:
1. Is it just the preverifier that is the issue, or are there other 
components we need to worry about as well?
        
Given projects like MPowerPlayer and microemu, Mac has basic 
emulation
      
functionality.  These emulators are limited in their JSR support, but 
are probably "good enough" for now.  The problem does tend to be with 
the preverifier.
      
2. Should we ask Sun if the existing PhoneME preverifier can also be 
released as EPL code within MTJ?
        
You can ask, but my guess is that they will say that it is already 
"open source".  The problem is that it is GPL.
      
My thought is we must have Windows, Linux, and Mac support in MTJ, 
the real question is the best way to get there (fast).....
        
I agree.  I did some experimentation and found that building the 
preverifier from source for the Mac was not terribly difficult.  This 
might be a reasonable starting point... I can always provide it as a 
download for people to avoid EPL licensing issues with the GPL code.
To make user of this, it would be necessary to allow an arbitrary 
external preverifier to be associated with mpowerplayer and microemu 
projects.  This is not supported by the EclipseME code on which MTJ 
is
      
based.  The best thing we could do would be to allow the user to 
choose an external preverifier or the built-in preverifier for *any* 
device.
In terms of the built-in preverifier, I had a quick look at ASM 3.0 
in
      
the process of doing something else for work.  It appears there are 
actually some better API's in there that could be used to "fix up"
the
      
current implementation.  Unfortunately, that may not be doable in the 
short term.
In summary... I think there are some reasonable options for Mac, but 
one of them is going to force the user through more hoops and the 
other requires more work on MTJ part.
Craig
_______________________________________________
dsdp-mtj-dev mailing list
dsdp-mtj-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/dsdp-mtj-dev
_______________________________________________
dsdp-mtj-dev mailing list
dsdp-mtj-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/dsdp-mtj-dev
      
_______________________________________________
dsdp-mtj-dev mailing list
dsdp-mtj-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/dsdp-mtj-dev
_______________________________________________
dsdp-mtj-dev mailing list
dsdp-mtj-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/dsdp-mtj-dev
_______________________________________________
dsdp-mtj-dev mailing list
dsdp-mtj-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/dsdp-mtj-dev
    
_______________________________________________
dsdp-mtj-dev mailing list
dsdp-mtj-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/dsdp-mtj-dev
_______________________________________________
dsdp-mtj-dev mailing list
dsdp-mtj-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/dsdp-mtj-dev