harmony-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Alexei Fedotov <alexei.fedo...@gmail.com>
Subject Re: Google Summer of Code 2009
Date Wed, 18 Mar 2009 14:48:40 GMT
Jimmy,

As for your questions, I have a strong belief that a mentor should
have 90% of a proper GSoC project design before suggesting the project
to students even if students are supposed to design the whole project.
I think there is still place for analysis and thinking about your
projects .

Good luck.


2009/3/18 Jimmy,Jing Lv <firepure@gmail.com>:
> Hi Alexier,
>
>     Let me explain a little more. Basically, the motivation comes
> from the requirement of the Apache Harmony Projects.
>
> 1. the requirement of the localized messages
>    Indeed we are lack of them, and we need them for customer use.
> However we see there's no resource, no translation team for Harmony
> Project to do this. And of course for GSoC, it is impossible for it to
> support a project simply translate the message. But an automatic tool
> sounds reasonable.
>    Technically, yes we have eclipse or some other tools to extract
> the String, but I don't see there's a full tool, automatically extract
> the messages, analysis if necessary to translate, translate and
> adjust, and at last inject into the source code with ResourceBundle or
> property loading mechanism and apply the localization.
>    This, I believe, is a Win-Win project, help the Harmony project to
> be localized, as well as help student to train their skill, thus
> sounds acceptable for GSoC. If some other project will use it, that'll
> be a plus.
>
> 2. the new feature of Apache Harmony project
> a. the requirement of the smallest jre,
>     Yes I see FreeCiv - an amazing project. However as we see, not
> all project get to know if the jre can be customized, and not all
> customer know how to customize the jre. It may be an amazing tool for
> Harmony than Sun, help the customers to make up his smallest jre
> according his own project, which help them to make their program
> download size to be smallest. Do you think it'll be a nice feature for
> Harmony to our customers?
>
> b. the OSGi/updater
>     OSGi is hot and Apache Harmony has already make its modules
> bundle. I've heard of many java developers believe if the jdk can
> adopt OSGi is helpful. It will help some project to be OSGi without
> apply OSGi module on their projects. And what's more, yes I've got a
> little investigation on this topic as well, it seems Apache Felix can
> be applied on Harmony with a few lines changed and a few new methods
> enhanced on vm, and then start well, And eclipse Equinox looks
> adaptable as well. If we can go a little further, it may make the
> Harmony Project to be the first OSGi-ed JDK on the world - cool and
> attractive to the potential customers, isn't it? If so we can apply a
> update mechanism for Harmony easily.
>
>     Yes I see the new features are beyond the Spec, I think it's OK
> if we want to excel Sun's implementation and attract more customers.
> Do you think the motivation is acceptable? Is there any technical
> fault on these topics?
>
> 2009/3/18 Alexei Fedotov <alexei.fedotov@gmail.com>:
>> Hello Jimmy,
>> Being brief, I lack your motivation, not technical details.
>>
>> 1. I extracted minimal class sets using standard logging to enable
>> different applications on Harmony more than once. For example, this
>> was done for initial analysis of FreeCiv GSoC project. Why one would
>> need a tool?
>>
>> 2. Eclipse and many other tools have automatic string extractors for
>> further localization. I believe you cannot miss all these
>> //$NON-NLS-1$ comments in the code. Your proposal lacks explanation
>> why other open source tools do not fit.
>>
>> 3. I think that before one develops a Harmony updater, a research is
>> to be conducted why existing updaters cannot be adopted. Yes, I truly
>> believe that an updater is a separate project unless it complies with
>> web start specification or OSGi updater specification. For OSGi, one
>> should research if OSGi code can be simply adopted instead of
>> rewritten.
>>
>> 4. I expect some words about isolation and OPEN component management
>> as a motivation for multi-vm. IMHO, this task, if done correctly, is
>> impossible for a student.
>>
>> Thanks!
>>
>>
>> 2009/3/18 Jimmy,Jing Lv <firepure@gmail.com>:
>>> Hi,
>>>
>>>     Thanks Alexei for your suggestion! Sorry maybe I make the
>>> abstract on the wiki too short to be understandable, I will enrich the
>>> information of the project motivation.
>>>
>>>     The translation tools as I mentioned there, was not focus on the
>>> translation(as we know there's so many translation tools) but the
>>> automatic fetching/injecting mechanism of the open source project
>>> localizable messages. The first motivation was that we Harmony Project
>>> lack those messages compared to Sun's implementation thus was not
>>> friendly to our users. However think deeper we see many open source
>>> projects suffer the similar problem. So the tools was planned to
>>> automatically help Harmony the message with this tool, what's more,
>>> may help some other open source project, especially java projects. So
>>> I believe this is still closely related to Harmony and is a
>>> requirement of the project(not sure if it should be titled with
>>> "Harmony-tool-5"?)
>>>
>>>     The other projects are similar, comes from the requirement of
>>> harmony project and customers requirements. Currently the Harmony
>>> project lack a automatic updater, NSIS offer only windows installation
>>> tool. The basic idea was simple, create a version checker/updater for
>>> Harmony. The smallest classes selector focus on customized-harmony-jre
>>> which may be friendly to customer usage, to find the classes with
>>> -verbose was easy(but still may not be enough, as we don't really know
>>> if the current set cover all classes, maybe some classes will be only
>>> load at some special case), but the problem we want to solve is the
>>> automatic select/build/test/packet for customer application with a
>>> full but smallest JRE, which may be valuable to customer (e.g, eclipse
>>> does not need to customize our jre with much effort).
>>>
>>>     I don't think they can be called as "new open source project",
>>> just some tool set for harmony. What do you think?
>>>
>>>     There's another thought, we may enable Apache Harmony JDK with
>>> OSGi feature. Yes it was beyond the spec, however it seems valuable if
>>> we find out a light way, at least we see Harmony modules are already
>>> bundled.
>>>
>>> 2009/3/17 Alexei Fedotov <alexei.fedotov@gmail.com>:
>>>> I like GSoC tasks from Sean and Andrew.
>>>>
>>>> Jimmy,
>>>> Can you provide a sort of analysis of your tasks? Are there any open
>>>> source automatic translation tools? How a web updater relates to NSIS
>>>> and Java WebStart? Is there any relation between your multi-vm and
>>>> isolation API? What is in a smallest class set except -verbose:class?
>>>> Please, provide more motivation.
>>>>
>>>> As one of Apache gurus said, "You want start a new open source project? Don't."
>>>>
>>>> Thanks!
>>>>
>>>> On Tue, Mar 17, 2009 at 2:44 PM, Jimmy,Jing Lv <firepure@gmail.com>
wrote:
>>>>> Hi,
>>>>>
>>>>>    I've added 4 proposal there. Looking for discussions/suggestions/comments
:)
>>>>>
>>>>> 2009/3/16 Sian January <sianjanuary@googlemail.com>:
>>>>>> Just wanted to encourage people to write their ideas up on the Wiki
-
>>>>>> only Oliver has done it so far and there is a deadline.
>>>>>>
>>>>>> Thanks,
>>>>>>
>>>>>> Sian
>>>>>>
>>>>>>
>>>>>> 2009/3/12 Andrew Cornwall <andrew.pack200@gmail.com>:
>>>>>>> Is there any desire to extend VMTT?
>>>>>>>
>>>>>>>  - add a real assembly format (jsr :label or something like
that, instead of
>>>>>>> jsr -11.) (Does jasm do this, and if so could we use their format?)
>>>>>>>  - add flexibility for bad classes
>>>>>>>    - specifying the padding for tableswitch/lookupswitch
>>>>>>>    - allow double/long constantpool entries without subsequent
constant
>>>>>>> pool entry
>>>>>>>    - mixed asm and bin in methods
>>>>>>>    - allow insertion of binary data at other places (eg constant
pool)
>>>>>>>  - fix bugs
>>>>>>>
>>>>>>> Just some thoughts...
>>>>>>>
>>>>>>>    Andrew Jr.
>>>>>>>
>>>>>>> On Wed, Mar 4, 2009 at 5:03 PM, Xiao-Feng Li <xiaofeng.li@gmail.com>
wrote:
>>>>>>>
>>>>>>>> On Wed, Mar 4, 2009 at 10:43 PM, Egor Pasko <egor.pasko@gmail.com>
wrote:
>>>>>>>> > On the 0x56A day of Apache Harmony Xiao-Feng Li wrote:
>>>>>>>> >> On Wed, Mar 4, 2009 at 6:46 PM, Egor Pasko <egor.pasko@gmail.com>
>>>>>>>> wrote:
>>>>>>>> >>> On the 0x56A day of Apache Harmony Xiao-Feng
Li wrote:
>>>>>>>> >>>> This is interesting. Project ideas related
to this include:
>>>>>>>> >>>>
>>>>>>>> >>>> 1. Make Apache Harmony support Google Android
applications. (Android
>>>>>>>> >>>> on Harmony should be much faster with the
much more powerful JIT and
>>>>>>>> >>>> GC).
>>>>>>>> >>>
>>>>>>>> >>> Do you mean replacing Dalvik on top of the Android
stack or making a
>>>>>>>> >>> system that works on a common desktop?
>>>>>>>> >>
>>>>>>>> >> Something like that... :)  At the moment I prefer
the second approach:
>>>>>>>> >> to make Harmony ready for Android applications on
desktop. The first
>>>>>>>> >> approach probably should be a project of Google
Android.
>>>>>>>> >
>>>>>>>> > sounds like a nice idea :) though I cannot imagine how
much work is
>>>>>>>> > required to make this happen..
>>>>>>>>
>>>>>>>> Yes, there are lots of work in it. Well only when somebody
really
>>>>>>>> starts thinking about it,  can we gradually get some good
solution. To
>>>>>>>> replace Dalvik with Harmony in Android stack might be easier.
In any
>>>>>>>> case, it requires to run Android on top of a desktop OS,
with full
>>>>>>>> libc and utils, because that's Harmony needs. That's why
I have the
>>>>>>>> second project idea, to reduce Harmony into a very concise
version
>>>>>>>> that requires only minimum OS supports, as minimum as Android
>>>>>>>> requires. :)
>>>>>>>>
>>>>>>>> Thanks,
>>>>>>>> xiaofeng
>>>>>>>>
>>>>>>>> > --
>>>>>>>> > Egor Pasko
>>>>>>>> >
>>>>>>>> >
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> --
>>>>>>>> Managed Runtime Technology Center, Intel
>>>>>>>>
>>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> --
>>>>>> Unless stated otherwise above:
>>>>>> IBM United Kingdom Limited - Registered in England and Wales with
number 741598.
>>>>>> Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire
PO6 3AU
>>>>>>
>>>>>
>>>>>
>>>>>
>>>>> --
>>>>>
>>>>> Best Regards!
>>>>>
>>>>> Jimmy, Jing Lv
>>>>> China Software Development Lab, IBM
>>>>>
>>>>
>>>>
>>>>
>>>> --
>>>> С уважением,
>>>> Алексей Федотов,
>>>> http://people.apache.org/~aaf/
>>>>
>>>
>>>
>>>
>>> --
>>>
>>> Best Regards!
>>>
>>> Jimmy, Jing Lv
>>> China Software Development Lab, IBM
>>>
>>
>>
>>
>> --
>> С уважением,
>> Алексей Федотов,
>> http://people.apache.org/~aaf/
>>
>
>
>
> --
>
> Best Regards!
>
> Jimmy, Jing Lv
> China Software Development Lab, IBM
>



-- 
С уважением,
Алексей Федотов,
http://people.apache.org/~aaf/

Mime
View raw message