You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@harmony.apache.org by Tony Wu <wu...@gmail.com> on 2007/03/28 08:29:11 UTC
[classlib]proposal: java6 kick off!
Hi all,
It's close to the end of Q1 2007. Is it time for us to consider the
JAVA SE 6 new features? IIRC it was written in our high level road
map. We could start with creating a branch and recording what should
we do and the progress on harmony wiki.
I did a quick look on luni and found there are many things to do:
java.lang
new methods of Math,
trivial update of String, System, depends on IO
java.util
Many misc improvement for common utilities,
significant improvement: Deque, NavigatableMap/Set;
ResourceBundle$Control, ServiceLoader
java.net
NetworkInterface enhancement, IDN, IRI, Cookie Manager, NTLM and HTTP
negotiate authentication, http server-jetty?
java.io
Console may be complex, ObjectStreamClass new method, not sure yet,
10 new methods of File, need native support
I really can't help going to implementing them ;-)
Anyone has interest? What's your opinion?
--
Tony Wu
China Software Development Lab, IBM
Re: [classlib]proposal: java6 kick off!
Posted by Alexey Petrenko <al...@gmail.com>.
Yep, looks like we need a branch...
SY, Alexey
2007/3/28, Tony Wu <wu...@gmail.com>:
> Hi all,
> It's close to the end of Q1 2007. Is it time for us to consider the
> JAVA SE 6 new features? IIRC it was written in our high level road
> map. We could start with creating a branch and recording what should
> we do and the progress on harmony wiki.
>
> I did a quick look on luni and found there are many things to do:
> java.lang
> new methods of Math,
> trivial update of String, System, depends on IO
>
> java.util
> Many misc improvement for common utilities,
> significant improvement: Deque, NavigatableMap/Set;
> ResourceBundle$Control, ServiceLoader
>
> java.net
> NetworkInterface enhancement, IDN, IRI, Cookie Manager, NTLM and HTTP
> negotiate authentication, http server-jetty?
>
> java.io
> Console may be complex, ObjectStreamClass new method, not sure yet,
> 10 new methods of File, need native support
>
> I really can't help going to implementing them ;-)
> Anyone has interest? What's your opinion?
>
> --
> Tony Wu
> China Software Development Lab, IBM
>
Re: [classlib]proposal: java6 kick off!
Posted by Tony Wu <wu...@gmail.com>.
Oh sure! Any help is welcome :)
There are many things to do in vm, such as parallel gc and some probes
api, which must be used in java.lang.management I think.
On 3/29/07, Mikhail Fursov <mi...@gmail.com> wrote:
> Tony,
> Do you need any special support from JIT or VM modules?
>
>
> On 3/29/07, Tony Wu <wu...@gmail.com> wrote:
> >
> > IMHO, call a java5 with extension as java5 is not a good idea. User
> > always consider the jdk he got is version 5, the new feature of 6 will
> > never be used but we still have to pay for them in bigger file size,
> > possible worse performance and extra potential risk.
> >
> > On 3/28/07, Mikhail Fursov <mi...@gmail.com> wrote:
> > > I agree that branching is good.
> > > My question was if Java6 is a simple Java5 extension then we can do all
> > > changes needed for Java6 in the trunk and call it Java5.
> > > But if we won't able to pass TCK scenarios this way - this is a good
> > reason
> > > to separate Java5/6 code.
> > >
> > > On 3/28/07, Tony Wu <wu...@gmail.com> wrote:
> > > >
> > > > Hi Mikhail,
> > > >
> > > > branching will bring many benefits,
> > > > developing work on branch will not affect the trunk. We can not assume
> > > > that all the middle version from 5 to 6 is stable with backward
> > > > compatibility. say, some spec changed.
> > > >
> > > > As Paulex said, we have to deliver version 5 in the future.
> > > > when the java 5 code is stable, we can create a tag for version 5,
> > > > then merge java6 branch to the trunk when all the java6 work is
> > > > complete.
> > > >
> > > > Plus the branching operation of subversion is very light weight.
> > > >
> > > > On 3/28/07, Mikhail Fursov <mi...@gmail.com> wrote:
> > > > > Does it really worth branching?
> > > > > If there are no conflicts with Java5 API why not to work it in the
> > > > active
> > > > > trunk?
> > > > >
> > > > > On 3/28/07, Tony Wu <wu...@gmail.com> wrote:
> > > > > >
> > > > > > Hi all,
> > > > > > It's close to the end of Q1 2007. Is it time for us to consider
> > the
> > > > > > JAVA SE 6 new features? IIRC it was written in our high level road
> > > > > > map. We could start with creating a branch and recording what
> > should
> > > > > > we do and the progress on harmony wiki.
> > > > > >
> > > > > > I did a quick look on luni and found there are many things to do:
> > > > > > java.lang
> > > > > > new methods of Math,
> > > > > > trivial update of String, System, depends on IO
> > > > > >
> > > > > > java.util
> > > > > > Many misc improvement for common utilities,
> > > > > > significant improvement: Deque, NavigatableMap/Set;
> > > > > > ResourceBundle$Control, ServiceLoader
> > > > > >
> > > > > > java.net
> > > > > > NetworkInterface enhancement, IDN, IRI, Cookie Manager, NTLM and
> > HTTP
> > > > > > negotiate authentication, http server-jetty?
> > > > > >
> > > > > > java.io
> > > > > > Console may be complex, ObjectStreamClass new method, not sure
> > yet,
> > > > > > 10 new methods of File, need native support
> > > > > >
> > > > > > I really can't help going to implementing them ;-)
> > > > > > Anyone has interest? What's your opinion?
> > > > > >
> > > > > > --
> > > > > > Tony Wu
> > > > > > China Software Development Lab, IBM
> > > > > >
> > > > >
> > > > >
> > > > >
> > > > > --
> > > > > Mikhail Fursov
> > > > >
> > > >
> > > >
> > > > --
> > > > Tony Wu
> > > > China Software Development Lab, IBM
> > > >
> > >
> > >
> > >
> > > --
> > > Mikhail Fursov
> > >
> >
> >
> > --
> > Tony Wu
> > China Software Development Lab, IBM
> >
>
>
>
> --
> Mikhail Fursov
>
--
Tony Wu
China Software Development Lab, IBM
Re: [classlib]proposal: java6 kick off!
Posted by Mikhail Fursov <mi...@gmail.com>.
Tony,
Do you need any special support from JIT or VM modules?
On 3/29/07, Tony Wu <wu...@gmail.com> wrote:
>
> IMHO, call a java5 with extension as java5 is not a good idea. User
> always consider the jdk he got is version 5, the new feature of 6 will
> never be used but we still have to pay for them in bigger file size,
> possible worse performance and extra potential risk.
>
> On 3/28/07, Mikhail Fursov <mi...@gmail.com> wrote:
> > I agree that branching is good.
> > My question was if Java6 is a simple Java5 extension then we can do all
> > changes needed for Java6 in the trunk and call it Java5.
> > But if we won't able to pass TCK scenarios this way - this is a good
> reason
> > to separate Java5/6 code.
> >
> > On 3/28/07, Tony Wu <wu...@gmail.com> wrote:
> > >
> > > Hi Mikhail,
> > >
> > > branching will bring many benefits,
> > > developing work on branch will not affect the trunk. We can not assume
> > > that all the middle version from 5 to 6 is stable with backward
> > > compatibility. say, some spec changed.
> > >
> > > As Paulex said, we have to deliver version 5 in the future.
> > > when the java 5 code is stable, we can create a tag for version 5,
> > > then merge java6 branch to the trunk when all the java6 work is
> > > complete.
> > >
> > > Plus the branching operation of subversion is very light weight.
> > >
> > > On 3/28/07, Mikhail Fursov <mi...@gmail.com> wrote:
> > > > Does it really worth branching?
> > > > If there are no conflicts with Java5 API why not to work it in the
> > > active
> > > > trunk?
> > > >
> > > > On 3/28/07, Tony Wu <wu...@gmail.com> wrote:
> > > > >
> > > > > Hi all,
> > > > > It's close to the end of Q1 2007. Is it time for us to consider
> the
> > > > > JAVA SE 6 new features? IIRC it was written in our high level road
> > > > > map. We could start with creating a branch and recording what
> should
> > > > > we do and the progress on harmony wiki.
> > > > >
> > > > > I did a quick look on luni and found there are many things to do:
> > > > > java.lang
> > > > > new methods of Math,
> > > > > trivial update of String, System, depends on IO
> > > > >
> > > > > java.util
> > > > > Many misc improvement for common utilities,
> > > > > significant improvement: Deque, NavigatableMap/Set;
> > > > > ResourceBundle$Control, ServiceLoader
> > > > >
> > > > > java.net
> > > > > NetworkInterface enhancement, IDN, IRI, Cookie Manager, NTLM and
> HTTP
> > > > > negotiate authentication, http server-jetty?
> > > > >
> > > > > java.io
> > > > > Console may be complex, ObjectStreamClass new method, not sure
> yet,
> > > > > 10 new methods of File, need native support
> > > > >
> > > > > I really can't help going to implementing them ;-)
> > > > > Anyone has interest? What's your opinion?
> > > > >
> > > > > --
> > > > > Tony Wu
> > > > > China Software Development Lab, IBM
> > > > >
> > > >
> > > >
> > > >
> > > > --
> > > > Mikhail Fursov
> > > >
> > >
> > >
> > > --
> > > Tony Wu
> > > China Software Development Lab, IBM
> > >
> >
> >
> >
> > --
> > Mikhail Fursov
> >
>
>
> --
> Tony Wu
> China Software Development Lab, IBM
>
--
Mikhail Fursov
Re: [classlib]proposal: java6 kick off!
Posted by Tony Wu <wu...@gmail.com>.
IMHO, call a java5 with extension as java5 is not a good idea. User
always consider the jdk he got is version 5, the new feature of 6 will
never be used but we still have to pay for them in bigger file size,
possible worse performance and extra potential risk.
On 3/28/07, Mikhail Fursov <mi...@gmail.com> wrote:
> I agree that branching is good.
> My question was if Java6 is a simple Java5 extension then we can do all
> changes needed for Java6 in the trunk and call it Java5.
> But if we won't able to pass TCK scenarios this way - this is a good reason
> to separate Java5/6 code.
>
> On 3/28/07, Tony Wu <wu...@gmail.com> wrote:
> >
> > Hi Mikhail,
> >
> > branching will bring many benefits,
> > developing work on branch will not affect the trunk. We can not assume
> > that all the middle version from 5 to 6 is stable with backward
> > compatibility. say, some spec changed.
> >
> > As Paulex said, we have to deliver version 5 in the future.
> > when the java 5 code is stable, we can create a tag for version 5,
> > then merge java6 branch to the trunk when all the java6 work is
> > complete.
> >
> > Plus the branching operation of subversion is very light weight.
> >
> > On 3/28/07, Mikhail Fursov <mi...@gmail.com> wrote:
> > > Does it really worth branching?
> > > If there are no conflicts with Java5 API why not to work it in the
> > active
> > > trunk?
> > >
> > > On 3/28/07, Tony Wu <wu...@gmail.com> wrote:
> > > >
> > > > Hi all,
> > > > It's close to the end of Q1 2007. Is it time for us to consider the
> > > > JAVA SE 6 new features? IIRC it was written in our high level road
> > > > map. We could start with creating a branch and recording what should
> > > > we do and the progress on harmony wiki.
> > > >
> > > > I did a quick look on luni and found there are many things to do:
> > > > java.lang
> > > > new methods of Math,
> > > > trivial update of String, System, depends on IO
> > > >
> > > > java.util
> > > > Many misc improvement for common utilities,
> > > > significant improvement: Deque, NavigatableMap/Set;
> > > > ResourceBundle$Control, ServiceLoader
> > > >
> > > > java.net
> > > > NetworkInterface enhancement, IDN, IRI, Cookie Manager, NTLM and HTTP
> > > > negotiate authentication, http server-jetty?
> > > >
> > > > java.io
> > > > Console may be complex, ObjectStreamClass new method, not sure yet,
> > > > 10 new methods of File, need native support
> > > >
> > > > I really can't help going to implementing them ;-)
> > > > Anyone has interest? What's your opinion?
> > > >
> > > > --
> > > > Tony Wu
> > > > China Software Development Lab, IBM
> > > >
> > >
> > >
> > >
> > > --
> > > Mikhail Fursov
> > >
> >
> >
> > --
> > Tony Wu
> > China Software Development Lab, IBM
> >
>
>
>
> --
> Mikhail Fursov
>
--
Tony Wu
China Software Development Lab, IBM
Re: [classlib]proposal: java6 kick off!
Posted by Mikhail Fursov <mi...@gmail.com>.
I agree that branching is good.
My question was if Java6 is a simple Java5 extension then we can do all
changes needed for Java6 in the trunk and call it Java5.
But if we won't able to pass TCK scenarios this way - this is a good reason
to separate Java5/6 code.
On 3/28/07, Tony Wu <wu...@gmail.com> wrote:
>
> Hi Mikhail,
>
> branching will bring many benefits,
> developing work on branch will not affect the trunk. We can not assume
> that all the middle version from 5 to 6 is stable with backward
> compatibility. say, some spec changed.
>
> As Paulex said, we have to deliver version 5 in the future.
> when the java 5 code is stable, we can create a tag for version 5,
> then merge java6 branch to the trunk when all the java6 work is
> complete.
>
> Plus the branching operation of subversion is very light weight.
>
> On 3/28/07, Mikhail Fursov <mi...@gmail.com> wrote:
> > Does it really worth branching?
> > If there are no conflicts with Java5 API why not to work it in the
> active
> > trunk?
> >
> > On 3/28/07, Tony Wu <wu...@gmail.com> wrote:
> > >
> > > Hi all,
> > > It's close to the end of Q1 2007. Is it time for us to consider the
> > > JAVA SE 6 new features? IIRC it was written in our high level road
> > > map. We could start with creating a branch and recording what should
> > > we do and the progress on harmony wiki.
> > >
> > > I did a quick look on luni and found there are many things to do:
> > > java.lang
> > > new methods of Math,
> > > trivial update of String, System, depends on IO
> > >
> > > java.util
> > > Many misc improvement for common utilities,
> > > significant improvement: Deque, NavigatableMap/Set;
> > > ResourceBundle$Control, ServiceLoader
> > >
> > > java.net
> > > NetworkInterface enhancement, IDN, IRI, Cookie Manager, NTLM and HTTP
> > > negotiate authentication, http server-jetty?
> > >
> > > java.io
> > > Console may be complex, ObjectStreamClass new method, not sure yet,
> > > 10 new methods of File, need native support
> > >
> > > I really can't help going to implementing them ;-)
> > > Anyone has interest? What's your opinion?
> > >
> > > --
> > > Tony Wu
> > > China Software Development Lab, IBM
> > >
> >
> >
> >
> > --
> > Mikhail Fursov
> >
>
>
> --
> Tony Wu
> China Software Development Lab, IBM
>
--
Mikhail Fursov
Re: [classlib]proposal: java6 kick off!
Posted by Tony Wu <wu...@gmail.com>.
Hi Mikhail,
branching will bring many benefits,
developing work on branch will not affect the trunk. We can not assume
that all the middle version from 5 to 6 is stable with backward
compatibility. say, some spec changed.
As Paulex said, we have to deliver version 5 in the future.
when the java 5 code is stable, we can create a tag for version 5,
then merge java6 branch to the trunk when all the java6 work is
complete.
Plus the branching operation of subversion is very light weight.
On 3/28/07, Mikhail Fursov <mi...@gmail.com> wrote:
> Does it really worth branching?
> If there are no conflicts with Java5 API why not to work it in the active
> trunk?
>
> On 3/28/07, Tony Wu <wu...@gmail.com> wrote:
> >
> > Hi all,
> > It's close to the end of Q1 2007. Is it time for us to consider the
> > JAVA SE 6 new features? IIRC it was written in our high level road
> > map. We could start with creating a branch and recording what should
> > we do and the progress on harmony wiki.
> >
> > I did a quick look on luni and found there are many things to do:
> > java.lang
> > new methods of Math,
> > trivial update of String, System, depends on IO
> >
> > java.util
> > Many misc improvement for common utilities,
> > significant improvement: Deque, NavigatableMap/Set;
> > ResourceBundle$Control, ServiceLoader
> >
> > java.net
> > NetworkInterface enhancement, IDN, IRI, Cookie Manager, NTLM and HTTP
> > negotiate authentication, http server-jetty?
> >
> > java.io
> > Console may be complex, ObjectStreamClass new method, not sure yet,
> > 10 new methods of File, need native support
> >
> > I really can't help going to implementing them ;-)
> > Anyone has interest? What's your opinion?
> >
> > --
> > Tony Wu
> > China Software Development Lab, IBM
> >
>
>
>
> --
> Mikhail Fursov
>
--
Tony Wu
China Software Development Lab, IBM
Re: [classlib]proposal: java6 kick off!
Posted by Yang Paulex <pa...@gmail.com>.
2007/3/28, Mikhail Fursov <mi...@gmail.com>:
>
> Does it really worth branching?
> If there are no conflicts with Java5 API why not to work it in the active
> trunk?
Then how to export a Java 5 source tree?
I remembered there was discussion on the list, the proposals include tags in
source code, svn branch, etc. Maybe it's time to reactivate that thread and
try to get a solution on this.
I prefer branch a little, because tags definitely impair the readability of
source code. And if we introduced some preprocessor to create pure source
codes, the "merge back" after edit will be an issue.
The branch's main issue is also how to merge bug fixing/improvement between
branches, espcially our Java 5 trunk is still in progress and not stable
yet. Maybe we can get an daemon script to merge Java 5 branch modifications
to Java 6 branch on demand or daily, but some kind of manual work is
probably unavoidable.
On 3/28/07, Tony Wu <wu...@gmail.com> wrote:
> >
> > Hi all,
> > It's close to the end of Q1 2007. Is it time for us to consider the
> > JAVA SE 6 new features? IIRC it was written in our high level road
> > map. We could start with creating a branch and recording what should
> > we do and the progress on harmony wiki.
Not very sure this is the right time to go for Java 6, personally I prefer
we get a relative stable developer release at first, but I have no strong
feelings on this. Anyway, I think it will be interesting and has no harm to
create a wiki page on the gap between Java 5 and Java 6.
>
> > I did a quick look on luni and found there are many things to do:
> > java.lang
> > new methods of Math,
> > trivial update of String, System, depends on IO
> >
> > java.util
> > Many misc improvement for common utilities,
> > significant improvement: Deque, NavigatableMap/Set;
> > ResourceBundle$Control, ServiceLoader
> >
> > java.net
> > NetworkInterface enhancement, IDN, IRI, Cookie Manager, NTLM and HTTP
> > negotiate authentication, http server-jetty?
> >
> > java.io
> > Console may be complex, ObjectStreamClass new method, not sure yet,
> > 10 new methods of File, need native support
> >
> > I really can't help going to implementing them ;-)
> > Anyone has interest? What's your opinion?
> >
> > --
> > Tony Wu
> > China Software Development Lab, IBM
> >
>
>
>
> --
> Mikhail Fursov
>
--
Paulex Yang
China Software Development laboratory
IBM
Re: [classlib]proposal: java6 kick off!
Posted by Yang Paulex <pa...@gmail.com>.
2007/3/28, Alexey Varlamov <al...@gmail.com>:
>
> Suppose we're lucky and got TCK for Java5 in Q2, how would the new
> APIs affect TCK scores?
I think so, I will be suprised if TCK doesn't check public API signature.
On the other hand, I think we must get a way to release source and build for
Java 5(it's one of Harmony's main objectives anyway), whatever TCK license
status.
2007/3/28, Mikhail Fursov <mi...@gmail.com>:
> > Does it really worth branching?
> > If there are no conflicts with Java5 API why not to work it in the
> active
> > trunk?
> >
> > On 3/28/07, Tony Wu <wu...@gmail.com> wrote:
> > >
> > > Hi all,
> > > It's close to the end of Q1 2007. Is it time for us to consider the
> > > JAVA SE 6 new features? IIRC it was written in our high level road
> > > map. We could start with creating a branch and recording what should
> > > we do and the progress on harmony wiki.
> > >
> > > I did a quick look on luni and found there are many things to do:
> > > java.lang
> > > new methods of Math,
> > > trivial update of String, System, depends on IO
> > >
> > > java.util
> > > Many misc improvement for common utilities,
> > > significant improvement: Deque, NavigatableMap/Set;
> > > ResourceBundle$Control, ServiceLoader
> > >
> > > java.net
> > > NetworkInterface enhancement, IDN, IRI, Cookie Manager, NTLM and HTTP
> > > negotiate authentication, http server-jetty?
> > >
> > > java.io
> > > Console may be complex, ObjectStreamClass new method, not sure yet,
> > > 10 new methods of File, need native support
> > >
> > > I really can't help going to implementing them ;-)
> > > Anyone has interest? What's your opinion?
> > >
> > > --
> > > Tony Wu
> > > China Software Development Lab, IBM
> > >
> >
> >
> >
> > --
> > Mikhail Fursov
> >
>
--
Paulex Yang
China Software Development laboratory
IBM
Re: [classlib]proposal: java6 kick off!
Posted by Alexey Varlamov <al...@gmail.com>.
Suppose we're lucky and got TCK for Java5 in Q2, how would the new
APIs affect TCK scores?
2007/3/28, Mikhail Fursov <mi...@gmail.com>:
> Does it really worth branching?
> If there are no conflicts with Java5 API why not to work it in the active
> trunk?
>
> On 3/28/07, Tony Wu <wu...@gmail.com> wrote:
> >
> > Hi all,
> > It's close to the end of Q1 2007. Is it time for us to consider the
> > JAVA SE 6 new features? IIRC it was written in our high level road
> > map. We could start with creating a branch and recording what should
> > we do and the progress on harmony wiki.
> >
> > I did a quick look on luni and found there are many things to do:
> > java.lang
> > new methods of Math,
> > trivial update of String, System, depends on IO
> >
> > java.util
> > Many misc improvement for common utilities,
> > significant improvement: Deque, NavigatableMap/Set;
> > ResourceBundle$Control, ServiceLoader
> >
> > java.net
> > NetworkInterface enhancement, IDN, IRI, Cookie Manager, NTLM and HTTP
> > negotiate authentication, http server-jetty?
> >
> > java.io
> > Console may be complex, ObjectStreamClass new method, not sure yet,
> > 10 new methods of File, need native support
> >
> > I really can't help going to implementing them ;-)
> > Anyone has interest? What's your opinion?
> >
> > --
> > Tony Wu
> > China Software Development Lab, IBM
> >
>
>
>
> --
> Mikhail Fursov
>
Re: [classlib]proposal: java6 kick off!
Posted by Mikhail Fursov <mi...@gmail.com>.
Does it really worth branching?
If there are no conflicts with Java5 API why not to work it in the active
trunk?
On 3/28/07, Tony Wu <wu...@gmail.com> wrote:
>
> Hi all,
> It's close to the end of Q1 2007. Is it time for us to consider the
> JAVA SE 6 new features? IIRC it was written in our high level road
> map. We could start with creating a branch and recording what should
> we do and the progress on harmony wiki.
>
> I did a quick look on luni and found there are many things to do:
> java.lang
> new methods of Math,
> trivial update of String, System, depends on IO
>
> java.util
> Many misc improvement for common utilities,
> significant improvement: Deque, NavigatableMap/Set;
> ResourceBundle$Control, ServiceLoader
>
> java.net
> NetworkInterface enhancement, IDN, IRI, Cookie Manager, NTLM and HTTP
> negotiate authentication, http server-jetty?
>
> java.io
> Console may be complex, ObjectStreamClass new method, not sure yet,
> 10 new methods of File, need native support
>
> I really can't help going to implementing them ;-)
> Anyone has interest? What's your opinion?
>
> --
> Tony Wu
> China Software Development Lab, IBM
>
--
Mikhail Fursov