You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@activemq.apache.org by Mats Forslöf <Ma...@portwise.com> on 2006/05/30 15:44:53 UTC
The future of the ActiveMQ C++ clients
Hi all,
The openwire-cpp client is complete with one exception - the SSL part. The reason for this is that we would like it to support two-way SSL and before we starts to implement a two-way SSL socket we need to hear what you think about this regarding to the ActiveMQ server. Are there any problems with the current design to implement this and is it feasible for ActiveMQ to support it in the next release (4.1)? Also, should all clients (C, C++, .NET, Java) support this?
How do we progress with the AMQ C++ clients. Do we have a roadmap or any plans regarding the AMQ C++ clients? How do you think we should progress with the clients and into what? Should we merge the existing projects (OpenWire, STOMP, ...) or should they coexist? It is probably a bit confusing for the end users at the moment with several clients. I would like us to have a discussion on all this and more. Please, feel free to share your thoughts on the issues above.
Regards,
Mats Forslöf
RE: The future of the ActiveMQ C++ clients
Posted by ErinO <Er...@hotmail.com>.
Thanks Mats. We will let you guys know when it is done.
We just started it, and we are thinking make it very similar to Java
failover transport - adding some state classes, and the failover transport
will track those states and resend those info to broker in case of failure.
If you guys have any suggestions, feel free to drop a line.
Regards
Erin
Mats Forslöf wrote:
>
>
> Erin, a failover transport would be really cool, looking forward to a
> first glimpse. :) Please let me know if I we be of assistance in any way.
>
> It would be great if we all could unite and work on one client in the near
> future.
>
> Regards,
> Mats
>
> -----Original Message-----
> From: ErinO [mailto:ErinOceng@hotmail.com]
> Sent: den 1 juni 2006 01:48
> To: activemq-dev@geronimo.apache.org
> Subject: Re: The future of the ActiveMQ C++ clients
>
>
> Interesting, please pass along a snapshot.
>
> BTW, we are currently using Mat's CMS, and is trying to add failover
> transport to it. If your work can be easily combined with Mat's CMS work,
> that will be really cool.
>
> Erin
>
>
> Nathan Mittler wrote:
>>
>> Hey guys,
>> I've been working with a co-worker on a new c++ client that will
>> replace CMS and will truly support a pluggable protocol. We based
>> much of the code off of the .NET client, but had to do some
>> rearranging to get rid of protocol dependencies in the code. We're
>> nearing completion and have a full suite of CppUnit tests to go along
>> with it. The code base should be able to easily support the OpenWire
>> protocol as well (something CMS would have had some heartburn with).
>>
>> Up until now, Tim and I have just been working in secrecy off of our
>> own svn repo, since he doesn't have commit access to AMQ. But we
>> should have things to a decent state fairly soon (a couple of weeks,
>> perhaps). The first pass will only have STOMP implemented, but I was
>> hoping that once we get our stuff into Apache that you other guys
>> (Mats?) might give us a hand plugging in a connector for OpenWire?
>>
>> If anyone is interested, we can pass along a snapshot of our code base
>> so you can take a peek.
>>
>> Regards,
>> Nate
>>
>> On 5/30/06, Hiram Chirino <hi...@hiramchirino.com> wrote:
>>>
>>> Hi Mats,
>>>
>>> Yep, it should be possible to do 2 way ssl with ActiveMQ.. Perhaps
>>> the harder part will will finding a place to hold the client
>>> authentication info acquired from SSL to the broker. Perhaps we just
>>> stuff stuff into the connection info, not sure.
>>>
>>> I think it would be nice if all clients supported that feature, but I
>>> don't think it's a must have.
>>>
>>> As for the road map for the c++ clients, I'm hoping that things get
>>> consolidated down to 2 main sets of interfaces. 1st is the JMS like
>>> c++ API that supports all the features that the java client supports.
>>> 2nd is a simpler api that's perhaps only implemented using the stomp
>>> protocol.
>>>
>>> How we consolidate the APIs, I'm not sure. I think the C++ guys
>>> behind the different implementations are the ones that need to come
>>> to the table and discuss that. If they don't get consolidated, it
>>> might not be a bad thing. It gives users choice, and in the end they
>>> will pick what suits them best.
>>>
>>> Regards,
>>> Hiram
>>>
>>> On 5/30/06, Mats Forslöf <Ma...@portwise.com> wrote:
>>> > Hi all,
>>> >
>>> > The openwire-cpp client is complete with one exception - the SSL part.
>>> The reason for this is that we would like it to support two-way SSL
>>> and before we starts to implement a two-way SSL socket we need to
>>> hear what you think about this regarding to the ActiveMQ server. Are
>>> there any problems with the current design to implement this and is
>>> it feasible for ActiveMQ to support it in the next release (4.1)?
>>> Also, should all clients (C, C++, .NET, Java) support this?
>>> >
>>> > How do we progress with the AMQ C++ clients. Do we have a roadmap
>>> > or
>>> any
>>> plans regarding the AMQ C++ clients? How do you think we should
>>> progress with the clients and into what? Should we merge the existing
>>> projects (OpenWire, STOMP, ...) or should they coexist? It is
>>> probably a bit confusing for the end users at the moment with several
>>> clients. I would like us to have a discussion on all this and more.
>>> Please, feel free to share your thoughts on the issues above.
>>> >
>>> > Regards,
>>> > Mats Forslöf
>>> >
>>>
>>>
>>> --
>>> Regards,
>>> Hiram
>>>
>>
>>
>
> --
> View this message in context:
> http://www.nabble.com/The+future+of+the+ActiveMQ+C%2B%2B+clients-t1704829.html#a4653968
> Sent from the ActiveMQ - Dev forum at Nabble.com.
>
>
>
--
View this message in context: http://www.nabble.com/The+future+of+the+ActiveMQ+C%2B%2B+clients-t1704829.html#a4671633
Sent from the ActiveMQ - Dev forum at Nabble.com.
Re: The future of the ActiveMQ C++ clients
Posted by ErinO <Er...@hotmail.com>.
Interesting, please pass along a snapshot.
BTW, we are currently using Mat's CMS, and is trying to add failover
transport to it. If your work can be easily combined with Mat's CMS work,
that will be really cool.
Erin
Nathan Mittler wrote:
>
> Hey guys,
> I've been working with a co-worker on a new c++ client that will replace
> CMS
> and will truly support a pluggable protocol. We based much of the code
> off
> of the .NET client, but had to do some rearranging to get rid of protocol
> dependencies in the code. We're nearing completion and have a full suite
> of
> CppUnit tests to go along with it. The code base should be able to easily
> support the OpenWire protocol as well (something CMS would have had some
> heartburn with).
>
> Up until now, Tim and I have just been working in secrecy off of our own
> svn
> repo, since he doesn't have commit access to AMQ. But we should have
> things
> to a decent state fairly soon (a couple of weeks, perhaps). The first
> pass
> will only have STOMP implemented, but I was hoping that once we get our
> stuff into Apache that you other guys (Mats?) might give us a hand
> plugging
> in a connector for OpenWire?
>
> If anyone is interested, we can pass along a snapshot of our code base so
> you can take a peek.
>
> Regards,
> Nate
>
> On 5/30/06, Hiram Chirino <hi...@hiramchirino.com> wrote:
>>
>> Hi Mats,
>>
>> Yep, it should be possible to do 2 way ssl with ActiveMQ.. Perhaps
>> the harder part will will finding a place to hold the client
>> authentication info acquired from SSL to the broker. Perhaps we just
>> stuff stuff into the connection info, not sure.
>>
>> I think it would be nice if all clients supported that feature, but I
>> don't think it's a must have.
>>
>> As for the road map for the c++ clients, I'm hoping that things get
>> consolidated down to 2 main sets of interfaces. 1st is the JMS like
>> c++ API that supports all the features that the java client supports.
>> 2nd is a simpler api that's perhaps only implemented using the stomp
>> protocol.
>>
>> How we consolidate the APIs, I'm not sure. I think the C++ guys
>> behind the different implementations are the ones that need to come to
>> the table and discuss that. If they don't get consolidated, it might
>> not be a bad thing. It gives users choice, and in the end they will
>> pick what suits them best.
>>
>> Regards,
>> Hiram
>>
>> On 5/30/06, Mats Forslöf <Ma...@portwise.com> wrote:
>> > Hi all,
>> >
>> > The openwire-cpp client is complete with one exception - the SSL part.
>> The reason for this is that we would like it to support two-way SSL and
>> before we starts to implement a two-way SSL socket we need to hear what
>> you
>> think about this regarding to the ActiveMQ server. Are there any problems
>> with the current design to implement this and is it feasible for ActiveMQ
>> to
>> support it in the next release (4.1)? Also, should all clients (C, C++,
>> .NET, Java) support this?
>> >
>> > How do we progress with the AMQ C++ clients. Do we have a roadmap or
>> any
>> plans regarding the AMQ C++ clients? How do you think we should progress
>> with the clients and into what? Should we merge the existing projects
>> (OpenWire, STOMP, ...) or should they coexist? It is probably a bit
>> confusing for the end users at the moment with several clients. I would
>> like
>> us to have a discussion on all this and more. Please, feel free to share
>> your thoughts on the issues above.
>> >
>> > Regards,
>> > Mats Forslöf
>> >
>>
>>
>> --
>> Regards,
>> Hiram
>>
>
>
--
View this message in context: http://www.nabble.com/The+future+of+the+ActiveMQ+C%2B%2B+clients-t1704829.html#a4653968
Sent from the ActiveMQ - Dev forum at Nabble.com.
Re: The future of the ActiveMQ C++ clients
Posted by Nathan Mittler <na...@gmail.com>.
Hey guys,
I've been working with a co-worker on a new c++ client that will replace CMS
and will truly support a pluggable protocol. We based much of the code off
of the .NET client, but had to do some rearranging to get rid of protocol
dependencies in the code. We're nearing completion and have a full suite of
CppUnit tests to go along with it. The code base should be able to easily
support the OpenWire protocol as well (something CMS would have had some
heartburn with).
Up until now, Tim and I have just been working in secrecy off of our own svn
repo, since he doesn't have commit access to AMQ. But we should have things
to a decent state fairly soon (a couple of weeks, perhaps). The first pass
will only have STOMP implemented, but I was hoping that once we get our
stuff into Apache that you other guys (Mats?) might give us a hand plugging
in a connector for OpenWire?
If anyone is interested, we can pass along a snapshot of our code base so
you can take a peek.
Regards,
Nate
On 5/30/06, Hiram Chirino <hi...@hiramchirino.com> wrote:
>
> Hi Mats,
>
> Yep, it should be possible to do 2 way ssl with ActiveMQ.. Perhaps
> the harder part will will finding a place to hold the client
> authentication info acquired from SSL to the broker. Perhaps we just
> stuff stuff into the connection info, not sure.
>
> I think it would be nice if all clients supported that feature, but I
> don't think it's a must have.
>
> As for the road map for the c++ clients, I'm hoping that things get
> consolidated down to 2 main sets of interfaces. 1st is the JMS like
> c++ API that supports all the features that the java client supports.
> 2nd is a simpler api that's perhaps only implemented using the stomp
> protocol.
>
> How we consolidate the APIs, I'm not sure. I think the C++ guys
> behind the different implementations are the ones that need to come to
> the table and discuss that. If they don't get consolidated, it might
> not be a bad thing. It gives users choice, and in the end they will
> pick what suits them best.
>
> Regards,
> Hiram
>
> On 5/30/06, Mats Forslöf <Ma...@portwise.com> wrote:
> > Hi all,
> >
> > The openwire-cpp client is complete with one exception - the SSL part.
> The reason for this is that we would like it to support two-way SSL and
> before we starts to implement a two-way SSL socket we need to hear what you
> think about this regarding to the ActiveMQ server. Are there any problems
> with the current design to implement this and is it feasible for ActiveMQ to
> support it in the next release (4.1)? Also, should all clients (C, C++,
> .NET, Java) support this?
> >
> > How do we progress with the AMQ C++ clients. Do we have a roadmap or any
> plans regarding the AMQ C++ clients? How do you think we should progress
> with the clients and into what? Should we merge the existing projects
> (OpenWire, STOMP, ...) or should they coexist? It is probably a bit
> confusing for the end users at the moment with several clients. I would like
> us to have a discussion on all this and more. Please, feel free to share
> your thoughts on the issues above.
> >
> > Regards,
> > Mats Forslöf
> >
>
>
> --
> Regards,
> Hiram
>
Re: The future of the ActiveMQ C++ clients
Posted by Hiram Chirino <hi...@hiramchirino.com>.
Hi Mats,
Yep, it should be possible to do 2 way ssl with ActiveMQ.. Perhaps
the harder part will will finding a place to hold the client
authentication info acquired from SSL to the broker. Perhaps we just
stuff stuff into the connection info, not sure.
I think it would be nice if all clients supported that feature, but I
don't think it's a must have.
As for the road map for the c++ clients, I'm hoping that things get
consolidated down to 2 main sets of interfaces. 1st is the JMS like
c++ API that supports all the features that the java client supports.
2nd is a simpler api that's perhaps only implemented using the stomp
protocol.
How we consolidate the APIs, I'm not sure. I think the C++ guys
behind the different implementations are the ones that need to come to
the table and discuss that. If they don't get consolidated, it might
not be a bad thing. It gives users choice, and in the end they will
pick what suits them best.
Regards,
Hiram
On 5/30/06, Mats Forslöf <Ma...@portwise.com> wrote:
> Hi all,
>
> The openwire-cpp client is complete with one exception - the SSL part. The reason for this is that we would like it to support two-way SSL and before we starts to implement a two-way SSL socket we need to hear what you think about this regarding to the ActiveMQ server. Are there any problems with the current design to implement this and is it feasible for ActiveMQ to support it in the next release (4.1)? Also, should all clients (C, C++, .NET, Java) support this?
>
> How do we progress with the AMQ C++ clients. Do we have a roadmap or any plans regarding the AMQ C++ clients? How do you think we should progress with the clients and into what? Should we merge the existing projects (OpenWire, STOMP, ...) or should they coexist? It is probably a bit confusing for the end users at the moment with several clients. I would like us to have a discussion on all this and more. Please, feel free to share your thoughts on the issues above.
>
> Regards,
> Mats Forslöf
>
--
Regards,
Hiram