You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@directmemory.apache.org by Christoph Engelbert <no...@apache.org> on 2012/10/23 21:18:49 UTC
MemoryBuffer interface
Hey guys,
some time before I mentioned that it would be nice to have a real
buffer interface to against. The actual implementation only had
ByteBuffer when using non Unsafe MemoryAllocators.
I started to add a clean interface, derived from the nettys
ChannelBuffer, to be used as the main accesspoint to every memory
access no matter what the underlying access layer looks like.
At the moment I'm working against the GIT fork on GitHub and I'll
like to see your opinion and ideas about the MemoryBuffer interface
and the general idea.
The two important commits are:
https://github.com/noctarius/directmemory/commit/5b3cf11af0e71f5961b1bfcf69b10f3cb9388ff6
https://github.com/noctarius/directmemory/commit/05082a6aa2cac91bb2ab6e104837bb1431dae90d
Looking forward to your replies especially because I'm not yet sure
how the general way of new features is :-)
Cheers Chris
Re: MemoryBuffer interface
Posted by "Johannes.Lichtenberger" <Jo...@uni-konstanz.de>.
On 10/24/2012 10:10 PM, Christoph Engelbert wrote:
> Am 24.10.2012 22:08, schrieb Johannes.Lichtenberger:
>> On 10/24/2012 09:39 PM, Christoph Engelbert wrote:
>>> Am 24.10.2012 21:21, schrieb Johannes.Lichtenberger:
>>>> On 10/24/2012 09:03 PM, Christoph Engelbert wrote:
>>>>> Am 24.10.2012 21:01, schrieb Christoph Engelbert:
>>>>>> Am 24.10.2012 21:00, schrieb Olivier Lamy:
>>>>>>> 2012/10/24 Raffaele P. Guidi <ra...@gmail.com>:
>>>>>>>> Really, really good. Well, if all tests pass why not starting
>>>>>>>> pushing the
>>>>>>>> changes to svn? :-)
>>>>>>> +1 :-)
>>>>>> I guess there are no Unsafe unittests at the moment ;-) The other
>>>>>> tests already passed.
>>>>>
>>>>> PS: I think it would be good to use JUnit Parameterized tests that
>>>>> the same test is executed for every available MemoryManager /
>>>>> Allocator.
>>>>
>>>> BTW: This is really great in TestNG using the @DataProvider
>>>> annotation together with @Test(dataProvider = "foo").
>>>>
>>>
>>> JUnit has the same ability using @Parameters
>>> http://www.mkyong.com/unittest/junit-4-tutorial-6-parameterized-test/
>>> ;-)
>>
>> Great, didn't know about this annotation :-)
>>
>> According to [1] from the same guy, however it seems that the
>> TestNG parameters are more powerful (not restricted to primitives).
>>
>
> Not restricted to primitives ;-)
>
> @Parameters
> public static Collection<Object[]> data() {
> return Arrays.asList( new Object[][] {{new
> MemoryManagerServiceImpl<Object>()},{new
> UnsafeMemoryManagerServiceImpl<Object>()}} );
> }
Ok, thanks ;-)
Re: MemoryBuffer interface
Posted by Christoph Engelbert <no...@apache.org>.
Am 24.10.2012 22:08, schrieb Johannes.Lichtenberger:
> On 10/24/2012 09:39 PM, Christoph Engelbert wrote:
>> Am 24.10.2012 21:21, schrieb Johannes.Lichtenberger:
>>> On 10/24/2012 09:03 PM, Christoph Engelbert wrote:
>>>> Am 24.10.2012 21:01, schrieb Christoph Engelbert:
>>>>> Am 24.10.2012 21:00, schrieb Olivier Lamy:
>>>>>> 2012/10/24 Raffaele P. Guidi <ra...@gmail.com>:
>>>>>>> Really, really good. Well, if all tests pass why not starting
>>>>>>> pushing the
>>>>>>> changes to svn? :-)
>>>>>> +1 :-)
>>>>> I guess there are no Unsafe unittests at the moment ;-) The other
>>>>> tests already passed.
>>>>
>>>> PS: I think it would be good to use JUnit Parameterized tests that
>>>> the same test is executed for every available MemoryManager /
>>>> Allocator.
>>>
>>> BTW: This is really great in TestNG using the @DataProvider
>>> annotation together with @Test(dataProvider = "foo").
>>>
>>
>> JUnit has the same ability using @Parameters
>> http://www.mkyong.com/unittest/junit-4-tutorial-6-parameterized-test/
>> ;-)
>
> Great, didn't know about this annotation :-)
>
> According to [1] from the same guy, however it seems that the
> TestNG parameters are more powerful (not restricted to primitives).
>
Not restricted to primitives ;-)
@Parameters
public static Collection<Object[]> data() {
return Arrays.asList( new Object[][] {{new
MemoryManagerServiceImpl<Object>()},{new
UnsafeMemoryManagerServiceImpl<Object>()}} );
}
> kind regards,
> Johannes
>
> [1] http://www.mkyong.com/unittest/junit-4-vs-testng-comparison/
>
Re: MemoryBuffer interface
Posted by "Johannes.Lichtenberger" <Jo...@uni-konstanz.de>.
On 10/24/2012 09:39 PM, Christoph Engelbert wrote:
> Am 24.10.2012 21:21, schrieb Johannes.Lichtenberger:
>> On 10/24/2012 09:03 PM, Christoph Engelbert wrote:
>>> Am 24.10.2012 21:01, schrieb Christoph Engelbert:
>>>> Am 24.10.2012 21:00, schrieb Olivier Lamy:
>>>>> 2012/10/24 Raffaele P. Guidi <ra...@gmail.com>:
>>>>>> Really, really good. Well, if all tests pass why not starting
>>>>>> pushing the
>>>>>> changes to svn? :-)
>>>>> +1 :-)
>>>> I guess there are no Unsafe unittests at the moment ;-) The other
>>>> tests already passed.
>>>
>>> PS: I think it would be good to use JUnit Parameterized tests that
>>> the same test is executed for every available MemoryManager /
>>> Allocator.
>>
>> BTW: This is really great in TestNG using the @DataProvider
>> annotation together with @Test(dataProvider = "foo").
>>
>
> JUnit has the same ability using @Parameters
> http://www.mkyong.com/unittest/junit-4-tutorial-6-parameterized-test/ ;-)
Great, didn't know about this annotation :-)
According to [1] from the same guy, however it seems that the TestNG
parameters are more powerful (not restricted to primitives).
kind regards,
Johannes
[1] http://www.mkyong.com/unittest/junit-4-vs-testng-comparison/
Re: MemoryBuffer interface
Posted by Christoph Engelbert <no...@apache.org>.
Am 24.10.2012 21:21, schrieb Johannes.Lichtenberger:
> On 10/24/2012 09:03 PM, Christoph Engelbert wrote:
>> Am 24.10.2012 21:01, schrieb Christoph Engelbert:
>>> Am 24.10.2012 21:00, schrieb Olivier Lamy:
>>>> 2012/10/24 Raffaele P. Guidi <ra...@gmail.com>:
>>>>> Really, really good. Well, if all tests pass why not starting
>>>>> pushing the
>>>>> changes to svn? :-)
>>>> +1 :-)
>>> I guess there are no Unsafe unittests at the moment ;-) The other
>>> tests already passed.
>>
>> PS: I think it would be good to use JUnit Parameterized tests that
>> the same test is executed for every available MemoryManager /
>> Allocator.
>
> BTW: This is really great in TestNG using the @DataProvider
> annotation together with @Test(dataProvider = "foo").
>
JUnit has the same ability using @Parameters
http://www.mkyong.com/unittest/junit-4-tutorial-6-parameterized-test/ ;-)
> kind regards,
> Johannes
>
Re: MemoryBuffer interface
Posted by "Johannes.Lichtenberger" <Jo...@uni-konstanz.de>.
On 10/24/2012 09:03 PM, Christoph Engelbert wrote:
> Am 24.10.2012 21:01, schrieb Christoph Engelbert:
>> Am 24.10.2012 21:00, schrieb Olivier Lamy:
>>> 2012/10/24 Raffaele P. Guidi <ra...@gmail.com>:
>>>> Really, really good. Well, if all tests pass why not starting pushing the
>>>> changes to svn? :-)
>>> +1 :-)
>> I guess there are no Unsafe unittests at the moment ;-) The other
>> tests already passed.
>
> PS: I think it would be good to use JUnit Parameterized tests that
> the same test is executed for every available MemoryManager / Allocator.
BTW: This is really great in TestNG using the @DataProvider annotation
together with @Test(dataProvider = "foo").
kind regards,
Johannes
Re: MemoryBuffer interface
Posted by Christoph Engelbert <no...@apache.org>.
Am 24.10.2012 21:01, schrieb Christoph Engelbert:
> Am 24.10.2012 21:00, schrieb Olivier Lamy:
>> 2012/10/24 Raffaele P. Guidi <ra...@gmail.com>:
>>> Really, really good. Well, if all tests pass why not starting pushing the
>>> changes to svn? :-)
>> +1 :-)
> I guess there are no Unsafe unittests at the moment ;-) The other
> tests already passed.
PS: I think it would be good to use JUnit Parameterized tests that
the same test is executed for every available MemoryManager / Allocator.
>
>>> Ciao,
>>> R
>>> Il giorno 24/ott/2012 11:35, "Christoph Engelbert" <no...@apache.org>
>>> ha scritto:
>>>
>>>> Hey,
>>>>
>>>> I added the codebase to support the existing UnsafeMemoryManager and
>>>> usage of the pointers.
>>>>
>>>> https://github.com/noctarius/directmemory/commit/dd666b673596c71bccf3d999da4da8c967370538
>>>>
>>>> Chris
>>>>
>>>> Am 24.10.2012 09:21, schrieb Raffaele P. Guidi:
>>>>> just put together a test using the UnsafeStore (there's already one
>>>>> available) and see how it works
>>>>>
>>>>> On Wed, Oct 24, 2012 at 6:51 AM, Christoph Engelbert
>>>>> <no...@apache.org>wrote:
>>>>>
>>>>>> Morning Raffaele,
>>>>>>
>>>>>> at the moment the store is not used but it should be easy to use the
>>>>>> pointers instead of a long for the memory address. I just need to
>>>>>> implement this.
>>>>>>
>>>>>> I also thought about some kind of a virtual memory file for swapping
>>>>>> purposes if the object should be just be removed from the cache but
>>>>>> wasn't used for a longer time (like the normal swap data).
>>>>>>
>>>>>> Cheers Chris
>>>>>>
>>>>>> Am 24.10.2012 00:41, schrieb Raffaele P. Guidi:
>>>>>>> Looks good - how does it play with the unsafe based store?
>>>>>>> Il giorno 23/ott/2012 21:21, "Christoph Engelbert" <
>>>> noctarius@apache.org
>>>>>>> ha scritto:
>>>>>>>
>>>>>>>> Hey guys,
>>>>>>>>
>>>>>>>> some time before I mentioned that it would be nice to have a real
>>>>>>>> buffer interface to against. The actual implementation only had
>>>>>>>> ByteBuffer when using non Unsafe MemoryAllocators.
>>>>>>>>
>>>>>>>> I started to add a clean interface, derived from the nettys
>>>>>>>> ChannelBuffer, to be used as the main accesspoint to every memory
>>>>>>>> access no matter what the underlying access layer looks like.
>>>>>>>>
>>>>>>>> At the moment I'm working against the GIT fork on GitHub and I'll
>>>>>>>> like to see your opinion and ideas about the MemoryBuffer interface
>>>>>>>> and the general idea.
>>>>>>>>
>>>>>>>> The two important commits are:
>>>>>>>>
>>>>>>>>
>>>> https://github.com/noctarius/directmemory/commit/5b3cf11af0e71f5961b1bfcf69b10f3cb9388ff6
>>>> https://github.com/noctarius/directmemory/commit/05082a6aa2cac91bb2ab6e104837bb1431dae90d
>>>>>>>> Looking forward to your replies especially because I'm not yet sure
>>>>>>>> how the general way of new features is :-)
>>>>>>>>
>>>>>>>> Cheers Chris
>>>>>>>>
>>
Re: MemoryBuffer interface
Posted by Christoph Engelbert <no...@apache.org>.
Am 24.10.2012 21:00, schrieb Olivier Lamy:
> 2012/10/24 Raffaele P. Guidi <ra...@gmail.com>:
>> Really, really good. Well, if all tests pass why not starting pushing the
>> changes to svn? :-)
> +1 :-)
I guess there are no Unsafe unittests at the moment ;-) The other
tests already passed.
>> Ciao,
>> R
>> Il giorno 24/ott/2012 11:35, "Christoph Engelbert" <no...@apache.org>
>> ha scritto:
>>
>>> Hey,
>>>
>>> I added the codebase to support the existing UnsafeMemoryManager and
>>> usage of the pointers.
>>>
>>> https://github.com/noctarius/directmemory/commit/dd666b673596c71bccf3d999da4da8c967370538
>>>
>>> Chris
>>>
>>> Am 24.10.2012 09:21, schrieb Raffaele P. Guidi:
>>>> just put together a test using the UnsafeStore (there's already one
>>>> available) and see how it works
>>>>
>>>> On Wed, Oct 24, 2012 at 6:51 AM, Christoph Engelbert
>>>> <no...@apache.org>wrote:
>>>>
>>>>> Morning Raffaele,
>>>>>
>>>>> at the moment the store is not used but it should be easy to use the
>>>>> pointers instead of a long for the memory address. I just need to
>>>>> implement this.
>>>>>
>>>>> I also thought about some kind of a virtual memory file for swapping
>>>>> purposes if the object should be just be removed from the cache but
>>>>> wasn't used for a longer time (like the normal swap data).
>>>>>
>>>>> Cheers Chris
>>>>>
>>>>> Am 24.10.2012 00:41, schrieb Raffaele P. Guidi:
>>>>>> Looks good - how does it play with the unsafe based store?
>>>>>> Il giorno 23/ott/2012 21:21, "Christoph Engelbert" <
>>> noctarius@apache.org
>>>>>> ha scritto:
>>>>>>
>>>>>>> Hey guys,
>>>>>>>
>>>>>>> some time before I mentioned that it would be nice to have a real
>>>>>>> buffer interface to against. The actual implementation only had
>>>>>>> ByteBuffer when using non Unsafe MemoryAllocators.
>>>>>>>
>>>>>>> I started to add a clean interface, derived from the nettys
>>>>>>> ChannelBuffer, to be used as the main accesspoint to every memory
>>>>>>> access no matter what the underlying access layer looks like.
>>>>>>>
>>>>>>> At the moment I'm working against the GIT fork on GitHub and I'll
>>>>>>> like to see your opinion and ideas about the MemoryBuffer interface
>>>>>>> and the general idea.
>>>>>>>
>>>>>>> The two important commits are:
>>>>>>>
>>>>>>>
>>> https://github.com/noctarius/directmemory/commit/5b3cf11af0e71f5961b1bfcf69b10f3cb9388ff6
>>> https://github.com/noctarius/directmemory/commit/05082a6aa2cac91bb2ab6e104837bb1431dae90d
>>>>>>> Looking forward to your replies especially because I'm not yet sure
>>>>>>> how the general way of new features is :-)
>>>>>>>
>>>>>>> Cheers Chris
>>>>>>>
>>>
>
>
Re: MemoryBuffer interface
Posted by Christoph Engelbert <no...@apache.org>.
Am 24.10.2012 23:09, schrieb Raffaele P. Guidi:
> Yay for the segfault! :-D
Ok first impression committed to the svn trunk. Tests are running
fine (at least for my local try) and I made some of the unittests
parameterized.
> Il giorno 24/ott/2012 23:04, "Christoph Engelbert" <no...@apache.org>
> ha scritto:
>
>> Am 24.10.2012 21:00, schrieb Olivier Lamy:
>>> 2012/10/24 Raffaele P. Guidi <ra...@gmail.com>:
>>>> Really, really good. Well, if all tests pass why not starting pushing
>> the
>>>> changes to svn? :-)
>>> +1 :-)
>> Ok back to topic I have added the JUnit extension and a lot of tests
>> are failing for the Unsafe implementation and I get a JVM SegFault
>> too :) I think there's something more to do. I'll commit it to the
>> SVN as far as the tests passing.
>>
>>>> Ciao,
>>>> R
>>>> Il giorno 24/ott/2012 11:35, "Christoph Engelbert" <
>> noctarius@apache.org>
>>>> ha scritto:
>>>>
>>>>> Hey,
>>>>>
>>>>> I added the codebase to support the existing UnsafeMemoryManager and
>>>>> usage of the pointers.
>>>>>
>>>>>
>> https://github.com/noctarius/directmemory/commit/dd666b673596c71bccf3d999da4da8c967370538
>>>>> Chris
>>>>>
>>>>> Am 24.10.2012 09:21, schrieb Raffaele P. Guidi:
>>>>>> just put together a test using the UnsafeStore (there's already one
>>>>>> available) and see how it works
>>>>>>
>>>>>> On Wed, Oct 24, 2012 at 6:51 AM, Christoph Engelbert
>>>>>> <no...@apache.org>wrote:
>>>>>>
>>>>>>> Morning Raffaele,
>>>>>>>
>>>>>>> at the moment the store is not used but it should be easy to use the
>>>>>>> pointers instead of a long for the memory address. I just need to
>>>>>>> implement this.
>>>>>>>
>>>>>>> I also thought about some kind of a virtual memory file for swapping
>>>>>>> purposes if the object should be just be removed from the cache but
>>>>>>> wasn't used for a longer time (like the normal swap data).
>>>>>>>
>>>>>>> Cheers Chris
>>>>>>>
>>>>>>> Am 24.10.2012 00:41, schrieb Raffaele P. Guidi:
>>>>>>>> Looks good - how does it play with the unsafe based store?
>>>>>>>> Il giorno 23/ott/2012 21:21, "Christoph Engelbert" <
>>>>> noctarius@apache.org
>>>>>>>> ha scritto:
>>>>>>>>
>>>>>>>>> Hey guys,
>>>>>>>>>
>>>>>>>>> some time before I mentioned that it would be nice to have a real
>>>>>>>>> buffer interface to against. The actual implementation only had
>>>>>>>>> ByteBuffer when using non Unsafe MemoryAllocators.
>>>>>>>>>
>>>>>>>>> I started to add a clean interface, derived from the nettys
>>>>>>>>> ChannelBuffer, to be used as the main accesspoint to every memory
>>>>>>>>> access no matter what the underlying access layer looks like.
>>>>>>>>>
>>>>>>>>> At the moment I'm working against the GIT fork on GitHub and I'll
>>>>>>>>> like to see your opinion and ideas about the MemoryBuffer interface
>>>>>>>>> and the general idea.
>>>>>>>>>
>>>>>>>>> The two important commits are:
>>>>>>>>>
>>>>>>>>>
>> https://github.com/noctarius/directmemory/commit/5b3cf11af0e71f5961b1bfcf69b10f3cb9388ff6
>> https://github.com/noctarius/directmemory/commit/05082a6aa2cac91bb2ab6e104837bb1431dae90d
>>>>>>>>> Looking forward to your replies especially because I'm not yet sure
>>>>>>>>> how the general way of new features is :-)
>>>>>>>>>
>>>>>>>>> Cheers Chris
>>>>>>>>>
>>>
>>
Re: MemoryBuffer interface
Posted by "Raffaele P. Guidi" <ra...@gmail.com>.
Yay for the segfault! :-D
Il giorno 24/ott/2012 23:04, "Christoph Engelbert" <no...@apache.org>
ha scritto:
> Am 24.10.2012 21:00, schrieb Olivier Lamy:
> > 2012/10/24 Raffaele P. Guidi <ra...@gmail.com>:
> >> Really, really good. Well, if all tests pass why not starting pushing
> the
> >> changes to svn? :-)
> > +1 :-)
>
> Ok back to topic I have added the JUnit extension and a lot of tests
> are failing for the Unsafe implementation and I get a JVM SegFault
> too :) I think there's something more to do. I'll commit it to the
> SVN as far as the tests passing.
>
> >> Ciao,
> >> R
> >> Il giorno 24/ott/2012 11:35, "Christoph Engelbert" <
> noctarius@apache.org>
> >> ha scritto:
> >>
> >>> Hey,
> >>>
> >>> I added the codebase to support the existing UnsafeMemoryManager and
> >>> usage of the pointers.
> >>>
> >>>
> https://github.com/noctarius/directmemory/commit/dd666b673596c71bccf3d999da4da8c967370538
> >>>
> >>> Chris
> >>>
> >>> Am 24.10.2012 09:21, schrieb Raffaele P. Guidi:
> >>>> just put together a test using the UnsafeStore (there's already one
> >>>> available) and see how it works
> >>>>
> >>>> On Wed, Oct 24, 2012 at 6:51 AM, Christoph Engelbert
> >>>> <no...@apache.org>wrote:
> >>>>
> >>>>> Morning Raffaele,
> >>>>>
> >>>>> at the moment the store is not used but it should be easy to use the
> >>>>> pointers instead of a long for the memory address. I just need to
> >>>>> implement this.
> >>>>>
> >>>>> I also thought about some kind of a virtual memory file for swapping
> >>>>> purposes if the object should be just be removed from the cache but
> >>>>> wasn't used for a longer time (like the normal swap data).
> >>>>>
> >>>>> Cheers Chris
> >>>>>
> >>>>> Am 24.10.2012 00:41, schrieb Raffaele P. Guidi:
> >>>>>> Looks good - how does it play with the unsafe based store?
> >>>>>> Il giorno 23/ott/2012 21:21, "Christoph Engelbert" <
> >>> noctarius@apache.org
> >>>>>> ha scritto:
> >>>>>>
> >>>>>>> Hey guys,
> >>>>>>>
> >>>>>>> some time before I mentioned that it would be nice to have a real
> >>>>>>> buffer interface to against. The actual implementation only had
> >>>>>>> ByteBuffer when using non Unsafe MemoryAllocators.
> >>>>>>>
> >>>>>>> I started to add a clean interface, derived from the nettys
> >>>>>>> ChannelBuffer, to be used as the main accesspoint to every memory
> >>>>>>> access no matter what the underlying access layer looks like.
> >>>>>>>
> >>>>>>> At the moment I'm working against the GIT fork on GitHub and I'll
> >>>>>>> like to see your opinion and ideas about the MemoryBuffer interface
> >>>>>>> and the general idea.
> >>>>>>>
> >>>>>>> The two important commits are:
> >>>>>>>
> >>>>>>>
> >>>
> https://github.com/noctarius/directmemory/commit/5b3cf11af0e71f5961b1bfcf69b10f3cb9388ff6
> >>>
> https://github.com/noctarius/directmemory/commit/05082a6aa2cac91bb2ab6e104837bb1431dae90d
> >>>>>>> Looking forward to your replies especially because I'm not yet sure
> >>>>>>> how the general way of new features is :-)
> >>>>>>>
> >>>>>>> Cheers Chris
> >>>>>>>
> >>>
> >
> >
>
>
Re: MemoryBuffer interface
Posted by Christoph Engelbert <no...@apache.org>.
Am 24.10.2012 21:00, schrieb Olivier Lamy:
> 2012/10/24 Raffaele P. Guidi <ra...@gmail.com>:
>> Really, really good. Well, if all tests pass why not starting pushing the
>> changes to svn? :-)
> +1 :-)
Ok back to topic I have added the JUnit extension and a lot of tests
are failing for the Unsafe implementation and I get a JVM SegFault
too :) I think there's something more to do. I'll commit it to the
SVN as far as the tests passing.
>> Ciao,
>> R
>> Il giorno 24/ott/2012 11:35, "Christoph Engelbert" <no...@apache.org>
>> ha scritto:
>>
>>> Hey,
>>>
>>> I added the codebase to support the existing UnsafeMemoryManager and
>>> usage of the pointers.
>>>
>>> https://github.com/noctarius/directmemory/commit/dd666b673596c71bccf3d999da4da8c967370538
>>>
>>> Chris
>>>
>>> Am 24.10.2012 09:21, schrieb Raffaele P. Guidi:
>>>> just put together a test using the UnsafeStore (there's already one
>>>> available) and see how it works
>>>>
>>>> On Wed, Oct 24, 2012 at 6:51 AM, Christoph Engelbert
>>>> <no...@apache.org>wrote:
>>>>
>>>>> Morning Raffaele,
>>>>>
>>>>> at the moment the store is not used but it should be easy to use the
>>>>> pointers instead of a long for the memory address. I just need to
>>>>> implement this.
>>>>>
>>>>> I also thought about some kind of a virtual memory file for swapping
>>>>> purposes if the object should be just be removed from the cache but
>>>>> wasn't used for a longer time (like the normal swap data).
>>>>>
>>>>> Cheers Chris
>>>>>
>>>>> Am 24.10.2012 00:41, schrieb Raffaele P. Guidi:
>>>>>> Looks good - how does it play with the unsafe based store?
>>>>>> Il giorno 23/ott/2012 21:21, "Christoph Engelbert" <
>>> noctarius@apache.org
>>>>>> ha scritto:
>>>>>>
>>>>>>> Hey guys,
>>>>>>>
>>>>>>> some time before I mentioned that it would be nice to have a real
>>>>>>> buffer interface to against. The actual implementation only had
>>>>>>> ByteBuffer when using non Unsafe MemoryAllocators.
>>>>>>>
>>>>>>> I started to add a clean interface, derived from the nettys
>>>>>>> ChannelBuffer, to be used as the main accesspoint to every memory
>>>>>>> access no matter what the underlying access layer looks like.
>>>>>>>
>>>>>>> At the moment I'm working against the GIT fork on GitHub and I'll
>>>>>>> like to see your opinion and ideas about the MemoryBuffer interface
>>>>>>> and the general idea.
>>>>>>>
>>>>>>> The two important commits are:
>>>>>>>
>>>>>>>
>>> https://github.com/noctarius/directmemory/commit/5b3cf11af0e71f5961b1bfcf69b10f3cb9388ff6
>>> https://github.com/noctarius/directmemory/commit/05082a6aa2cac91bb2ab6e104837bb1431dae90d
>>>>>>> Looking forward to your replies especially because I'm not yet sure
>>>>>>> how the general way of new features is :-)
>>>>>>>
>>>>>>> Cheers Chris
>>>>>>>
>>>
>
>
Re: MemoryBuffer interface
Posted by Olivier Lamy <ol...@apache.org>.
2012/10/24 Raffaele P. Guidi <ra...@gmail.com>:
> Really, really good. Well, if all tests pass why not starting pushing the
> changes to svn? :-)
+1 :-)
>
> Ciao,
> R
> Il giorno 24/ott/2012 11:35, "Christoph Engelbert" <no...@apache.org>
> ha scritto:
>
>> Hey,
>>
>> I added the codebase to support the existing UnsafeMemoryManager and
>> usage of the pointers.
>>
>> https://github.com/noctarius/directmemory/commit/dd666b673596c71bccf3d999da4da8c967370538
>>
>> Chris
>>
>> Am 24.10.2012 09:21, schrieb Raffaele P. Guidi:
>> > just put together a test using the UnsafeStore (there's already one
>> > available) and see how it works
>> >
>> > On Wed, Oct 24, 2012 at 6:51 AM, Christoph Engelbert
>> > <no...@apache.org>wrote:
>> >
>> >> Morning Raffaele,
>> >>
>> >> at the moment the store is not used but it should be easy to use the
>> >> pointers instead of a long for the memory address. I just need to
>> >> implement this.
>> >>
>> >> I also thought about some kind of a virtual memory file for swapping
>> >> purposes if the object should be just be removed from the cache but
>> >> wasn't used for a longer time (like the normal swap data).
>> >>
>> >> Cheers Chris
>> >>
>> >> Am 24.10.2012 00:41, schrieb Raffaele P. Guidi:
>> >>> Looks good - how does it play with the unsafe based store?
>> >>> Il giorno 23/ott/2012 21:21, "Christoph Engelbert" <
>> noctarius@apache.org
>> >>>
>> >>> ha scritto:
>> >>>
>> >>>> Hey guys,
>> >>>>
>> >>>> some time before I mentioned that it would be nice to have a real
>> >>>> buffer interface to against. The actual implementation only had
>> >>>> ByteBuffer when using non Unsafe MemoryAllocators.
>> >>>>
>> >>>> I started to add a clean interface, derived from the nettys
>> >>>> ChannelBuffer, to be used as the main accesspoint to every memory
>> >>>> access no matter what the underlying access layer looks like.
>> >>>>
>> >>>> At the moment I'm working against the GIT fork on GitHub and I'll
>> >>>> like to see your opinion and ideas about the MemoryBuffer interface
>> >>>> and the general idea.
>> >>>>
>> >>>> The two important commits are:
>> >>>>
>> >>>>
>> >>
>> https://github.com/noctarius/directmemory/commit/5b3cf11af0e71f5961b1bfcf69b10f3cb9388ff6
>> >>>>
>> >>
>> https://github.com/noctarius/directmemory/commit/05082a6aa2cac91bb2ab6e104837bb1431dae90d
>> >>>> Looking forward to your replies especially because I'm not yet sure
>> >>>> how the general way of new features is :-)
>> >>>>
>> >>>> Cheers Chris
>> >>>>
>> >>
>>
>>
--
Olivier Lamy
Talend: http://coders.talend.com
http://twitter.com/olamy | http://linkedin.com/in/olamy
Re: MemoryBuffer interface
Posted by "Raffaele P. Guidi" <ra...@gmail.com>.
Really, really good. Well, if all tests pass why not starting pushing the
changes to svn? :-)
Ciao,
R
Il giorno 24/ott/2012 11:35, "Christoph Engelbert" <no...@apache.org>
ha scritto:
> Hey,
>
> I added the codebase to support the existing UnsafeMemoryManager and
> usage of the pointers.
>
> https://github.com/noctarius/directmemory/commit/dd666b673596c71bccf3d999da4da8c967370538
>
> Chris
>
> Am 24.10.2012 09:21, schrieb Raffaele P. Guidi:
> > just put together a test using the UnsafeStore (there's already one
> > available) and see how it works
> >
> > On Wed, Oct 24, 2012 at 6:51 AM, Christoph Engelbert
> > <no...@apache.org>wrote:
> >
> >> Morning Raffaele,
> >>
> >> at the moment the store is not used but it should be easy to use the
> >> pointers instead of a long for the memory address. I just need to
> >> implement this.
> >>
> >> I also thought about some kind of a virtual memory file for swapping
> >> purposes if the object should be just be removed from the cache but
> >> wasn't used for a longer time (like the normal swap data).
> >>
> >> Cheers Chris
> >>
> >> Am 24.10.2012 00:41, schrieb Raffaele P. Guidi:
> >>> Looks good - how does it play with the unsafe based store?
> >>> Il giorno 23/ott/2012 21:21, "Christoph Engelbert" <
> noctarius@apache.org
> >>>
> >>> ha scritto:
> >>>
> >>>> Hey guys,
> >>>>
> >>>> some time before I mentioned that it would be nice to have a real
> >>>> buffer interface to against. The actual implementation only had
> >>>> ByteBuffer when using non Unsafe MemoryAllocators.
> >>>>
> >>>> I started to add a clean interface, derived from the nettys
> >>>> ChannelBuffer, to be used as the main accesspoint to every memory
> >>>> access no matter what the underlying access layer looks like.
> >>>>
> >>>> At the moment I'm working against the GIT fork on GitHub and I'll
> >>>> like to see your opinion and ideas about the MemoryBuffer interface
> >>>> and the general idea.
> >>>>
> >>>> The two important commits are:
> >>>>
> >>>>
> >>
> https://github.com/noctarius/directmemory/commit/5b3cf11af0e71f5961b1bfcf69b10f3cb9388ff6
> >>>>
> >>
> https://github.com/noctarius/directmemory/commit/05082a6aa2cac91bb2ab6e104837bb1431dae90d
> >>>> Looking forward to your replies especially because I'm not yet sure
> >>>> how the general way of new features is :-)
> >>>>
> >>>> Cheers Chris
> >>>>
> >>
>
>
Re: MemoryBuffer interface
Posted by Christoph Engelbert <no...@apache.org>.
Hey,
I added the codebase to support the existing UnsafeMemoryManager and
usage of the pointers.
https://github.com/noctarius/directmemory/commit/dd666b673596c71bccf3d999da4da8c967370538
Chris
Am 24.10.2012 09:21, schrieb Raffaele P. Guidi:
> just put together a test using the UnsafeStore (there's already one
> available) and see how it works
>
> On Wed, Oct 24, 2012 at 6:51 AM, Christoph Engelbert
> <no...@apache.org>wrote:
>
>> Morning Raffaele,
>>
>> at the moment the store is not used but it should be easy to use the
>> pointers instead of a long for the memory address. I just need to
>> implement this.
>>
>> I also thought about some kind of a virtual memory file for swapping
>> purposes if the object should be just be removed from the cache but
>> wasn't used for a longer time (like the normal swap data).
>>
>> Cheers Chris
>>
>> Am 24.10.2012 00:41, schrieb Raffaele P. Guidi:
>>> Looks good - how does it play with the unsafe based store?
>>> Il giorno 23/ott/2012 21:21, "Christoph Engelbert" <noctarius@apache.org
>>>
>>> ha scritto:
>>>
>>>> Hey guys,
>>>>
>>>> some time before I mentioned that it would be nice to have a real
>>>> buffer interface to against. The actual implementation only had
>>>> ByteBuffer when using non Unsafe MemoryAllocators.
>>>>
>>>> I started to add a clean interface, derived from the nettys
>>>> ChannelBuffer, to be used as the main accesspoint to every memory
>>>> access no matter what the underlying access layer looks like.
>>>>
>>>> At the moment I'm working against the GIT fork on GitHub and I'll
>>>> like to see your opinion and ideas about the MemoryBuffer interface
>>>> and the general idea.
>>>>
>>>> The two important commits are:
>>>>
>>>>
>> https://github.com/noctarius/directmemory/commit/5b3cf11af0e71f5961b1bfcf69b10f3cb9388ff6
>>>>
>> https://github.com/noctarius/directmemory/commit/05082a6aa2cac91bb2ab6e104837bb1431dae90d
>>>> Looking forward to your replies especially because I'm not yet sure
>>>> how the general way of new features is :-)
>>>>
>>>> Cheers Chris
>>>>
>>
Re: MemoryBuffer interface
Posted by "Raffaele P. Guidi" <ra...@gmail.com>.
just put together a test using the UnsafeStore (there's already one
available) and see how it works
On Wed, Oct 24, 2012 at 6:51 AM, Christoph Engelbert
<no...@apache.org>wrote:
> Morning Raffaele,
>
> at the moment the store is not used but it should be easy to use the
> pointers instead of a long for the memory address. I just need to
> implement this.
>
> I also thought about some kind of a virtual memory file for swapping
> purposes if the object should be just be removed from the cache but
> wasn't used for a longer time (like the normal swap data).
>
> Cheers Chris
>
> Am 24.10.2012 00:41, schrieb Raffaele P. Guidi:
> > Looks good - how does it play with the unsafe based store?
> > Il giorno 23/ott/2012 21:21, "Christoph Engelbert" <noctarius@apache.org
> >
> > ha scritto:
> >
> >> Hey guys,
> >>
> >> some time before I mentioned that it would be nice to have a real
> >> buffer interface to against. The actual implementation only had
> >> ByteBuffer when using non Unsafe MemoryAllocators.
> >>
> >> I started to add a clean interface, derived from the nettys
> >> ChannelBuffer, to be used as the main accesspoint to every memory
> >> access no matter what the underlying access layer looks like.
> >>
> >> At the moment I'm working against the GIT fork on GitHub and I'll
> >> like to see your opinion and ideas about the MemoryBuffer interface
> >> and the general idea.
> >>
> >> The two important commits are:
> >>
> >>
> https://github.com/noctarius/directmemory/commit/5b3cf11af0e71f5961b1bfcf69b10f3cb9388ff6
> >>
> >>
> https://github.com/noctarius/directmemory/commit/05082a6aa2cac91bb2ab6e104837bb1431dae90d
> >>
> >> Looking forward to your replies especially because I'm not yet sure
> >> how the general way of new features is :-)
> >>
> >> Cheers Chris
> >>
>
>
Re: MemoryBuffer interface
Posted by Christoph Engelbert <no...@apache.org>.
Morning Raffaele,
at the moment the store is not used but it should be easy to use the
pointers instead of a long for the memory address. I just need to
implement this.
I also thought about some kind of a virtual memory file for swapping
purposes if the object should be just be removed from the cache but
wasn't used for a longer time (like the normal swap data).
Cheers Chris
Am 24.10.2012 00:41, schrieb Raffaele P. Guidi:
> Looks good - how does it play with the unsafe based store?
> Il giorno 23/ott/2012 21:21, "Christoph Engelbert" <no...@apache.org>
> ha scritto:
>
>> Hey guys,
>>
>> some time before I mentioned that it would be nice to have a real
>> buffer interface to against. The actual implementation only had
>> ByteBuffer when using non Unsafe MemoryAllocators.
>>
>> I started to add a clean interface, derived from the nettys
>> ChannelBuffer, to be used as the main accesspoint to every memory
>> access no matter what the underlying access layer looks like.
>>
>> At the moment I'm working against the GIT fork on GitHub and I'll
>> like to see your opinion and ideas about the MemoryBuffer interface
>> and the general idea.
>>
>> The two important commits are:
>>
>> https://github.com/noctarius/directmemory/commit/5b3cf11af0e71f5961b1bfcf69b10f3cb9388ff6
>>
>> https://github.com/noctarius/directmemory/commit/05082a6aa2cac91bb2ab6e104837bb1431dae90d
>>
>> Looking forward to your replies especially because I'm not yet sure
>> how the general way of new features is :-)
>>
>> Cheers Chris
>>
Re: MemoryBuffer interface
Posted by "Raffaele P. Guidi" <ra...@gmail.com>.
Looks good - how does it play with the unsafe based store?
Il giorno 23/ott/2012 21:21, "Christoph Engelbert" <no...@apache.org>
ha scritto:
> Hey guys,
>
> some time before I mentioned that it would be nice to have a real
> buffer interface to against. The actual implementation only had
> ByteBuffer when using non Unsafe MemoryAllocators.
>
> I started to add a clean interface, derived from the nettys
> ChannelBuffer, to be used as the main accesspoint to every memory
> access no matter what the underlying access layer looks like.
>
> At the moment I'm working against the GIT fork on GitHub and I'll
> like to see your opinion and ideas about the MemoryBuffer interface
> and the general idea.
>
> The two important commits are:
>
> https://github.com/noctarius/directmemory/commit/5b3cf11af0e71f5961b1bfcf69b10f3cb9388ff6
>
> https://github.com/noctarius/directmemory/commit/05082a6aa2cac91bb2ab6e104837bb1431dae90d
>
> Looking forward to your replies especially because I'm not yet sure
> how the general way of new features is :-)
>
> Cheers Chris
>