You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@activemq.apache.org by "Jamie G." <ja...@gmail.com> on 2018/12/05 23:44:33 UTC
[Discuss] ActiveMQ CPP Client
Hi All,
I've started looking into a number of open cards on AMQ CPP client.
I've noticed that master is still 3.9.0-snapshot, when it likely
should be 3.10.0-SNAPSHOT now.
One supplied patch would bump openssl version to 1.0.2 from 0.9.8.
Would that bump be sufficient cause to increment the minor version?
(https://issues.apache.org/jira/browse/AMQCPP-619)
Once these patches are reviewed, and accepted - is there a release
process for the AMQ CPP client?
Tim, since you did a lot of work on this, your opinion would gladly be
appreciated.
Cheers,
Jamie
Re: [Discuss] ActiveMQ CPP Client
Posted by "Jamie G." <ja...@gmail.com>.
I've also merged https://issues.apache.org/jira/browse/AMQCPP-619 into
branch 3.10.x, taking care to ensure no tabs slipped into the patch :)
Cheers,
Jamie
On Fri, Dec 7, 2018 at 2:44 PM Jamie G. <ja...@gmail.com> wrote:
>
> I've created the 3.10.x branch based upon 3.9.x branch code.
>
> I'll start focusing patches towards that target.
>
> Cheers,
> Jamie
> On Fri, Dec 7, 2018 at 2:20 PM Timothy Bish <ta...@gmail.com> wrote:
> >
> > On 12/7/18 12:48 PM, Jamie G. wrote:
> > > Is that 3.10.x branch possibly local to your environment?
> > >
> > > I'm not seeing it on my pull...
> >
> > Ah, yes, never pushed, so should be good to go then on new branch
> >
> >
> > >
> > > On Fri, Dec 7, 2018 at 1:52 PM Timothy Bish <ta...@gmail.com> wrote:
> > >> On 12/7/18 12:01 PM, Jamie G. wrote:
> > >>> Thank you for the additional background information.
> > >>>
> > >>> So creating a 3.10.x branch off of current 3.9.x branch, than merging
> > >>> fixes/features to that branch would be acceptable?
> > >> I checked and there already is a 3.10.x branch which I must have been
> > >> intending to switch to at some point. You might be best to just drop
> > >> that branch and recut from current 3.9.x as that would be the best
> > >> starting point for patches. Once moved you might also want to check if
> > >> moving the supported APR lib versions up wasn't also a good idea given
> > >> the time between releases.
> > >>
> > >> I think there is some other work you might need to do which is that on
> > >> quite a few distros now the cppunit stuff won't build because they've
> > >> removed the package configuration script and gone to PKG configs so you
> > >> might want to test on more than one distro and or version # (I recall
> > >> some reports of problems on Fedora releases after 25 or 26)
> > >>
> > >> Also if you aren't testing on Windows you should build there and test as
> > >> well because that will often show you issues that need addressing before
> > >> release. Given the age of the code I'd be surprised if there weren't
> > >> some build and test issues that needed addressing.
> > >>
> > >>> Once a collection of patches have been applied to the 3.10.x, is there
> > >>> a release process that out lines how to get this published?
> > >>>
> > >>> http://activemq.apache.org/cms/creating-distributions.html seems to be
> > >>> pre-github.
> > >>>
> > >>> Cheers,
> > >>> Jamie
> > >>> On Fri, Dec 7, 2018 at 1:10 PM Timothy Bish <ta...@gmail.com> wrote:
> > >>>> My bad, I missed the CC. to dev on this so ignore my comment on keeping
> > >>>> in dev. :)
> > >>>>
> > >>>> On 12/7/18 11:17 AM, Jamie G. wrote:
> > >>>>> Hi Tim,
> > >>>>>
> > >>>>> Can i make master 3.10.0-SNAPSHOT or 4.0.0-SNAPSHOT? Which is preferred?
> > >>>>>
> > >>>>> In testing code on master the test suites passed on Linux & Mac OSX.
> > >>>>> Can you provide more details as to how master is not in a releasable
> > >>>>> state? Are there a collection of Jira cards reflecting what needs to
> > >>>>> be done to make Master releasable?
> > >>>>>
> > >>>>>
> > >>>>> On Fri, Dec 7, 2018 at 12:29 PM Timothy Bish <ta...@gmail.com> wrote:
> > >>>>>> On 12/5/18 6:44 PM, Jamie G. wrote:
> > >>>>>>> Hi All,
> > >>>>>>>
> > >>>>>>> I've started looking into a number of open cards on AMQ CPP client.
> > >>>>>>>
> > >>>>>>> I've noticed that master is still 3.9.0-snapshot, when it likely
> > >>>>>>> should be 3.10.0-SNAPSHOT now.
> > >>>>>> Master is not in a releasable state so all work you plan to do to create
> > >>>>>> patch releases should be directed to the patch branches unless you plan
> > >>>>>> on working on getting master into a releasable state which would be 4.0
> > >>>>>> but I doubt you plan to commit that much time to it so I'd suggest
> > >>>>>> planning on a 3.x release.
> > >>>>>>
> > >>>>>>
> > >>>>>>> One supplied patch would bump openssl version to 1.0.2 from 0.9.8.
> > >>>>>>> Would that bump be sufficient cause to increment the minor version?
> > >>>>>>> (https://issues.apache.org/jira/browse/AMQCPP-619)
> > >>>>>> 3.9.x is know to work on that older release so publishing a release in
> > >>>>>> that series would break the compatibility that would be expected from
> > >>>>>> that version range and so it would not be a drop in replacement. The
> > >>>>>> more sensible thing to do is to move onto a 3.10.x release series where
> > >>>>>> you break that know working range and state that it now requires OpenSSL
> > >>>>>> 1.0.x or whatever is the acceptable version. That way a hotfix release
> > >>>>>> of 3.9.x is still possible for anyone stuck on old system.
> > >>>>>>
> > >>>>>>
> > >>>>>>> Once these patches are reviewed, and accepted - is there a release
> > >>>>>>> process for the AMQ CPP client?
> > >>>>>> You would need to update autoconf versions numbers for the release,
> > >>>>>> updates code to reflect the release numbers and other things I'm
> > >>>>>> probably forgetting. Then you need to test on windows, linux and mac
> > >>>>>> with both SSL libs and no SSL libs installed to ensure the changes don't
> > >>>>>> break the build and test process. There are both unit tests and
> > >>>>>> integration tests that you need to get building and run to validate
> > >>>>>> things work on each platform.
> > >>>>>>
> > >>>>>> Once you validated the code works then you can you the mvn build to have
> > >>>>>> a set of archive built which you can then sign and checksum for release.
> > >>>>>>
> > >>>>>>> Tim, since you did a lot of work on this, your opinion would gladly be
> > >>>>>>> appreciated.
> > >>>>>>>
> > >>>>>>> Cheers,
> > >>>>>>> Jamie
> > >>>>>>>
> > >>>>>> --
> > >>>>>> Tim Bish
> > >>>>>>
> > >>>> --
> > >>>> Tim Bish
> > >>>>
> > >> --
> > >> Tim Bish
> > >>
> >
> > --
> > Tim Bish
> >
Re: [Discuss] ActiveMQ CPP Client
Posted by "Jamie G." <ja...@gmail.com>.
I've created the 3.10.x branch based upon 3.9.x branch code.
I'll start focusing patches towards that target.
Cheers,
Jamie
On Fri, Dec 7, 2018 at 2:20 PM Timothy Bish <ta...@gmail.com> wrote:
>
> On 12/7/18 12:48 PM, Jamie G. wrote:
> > Is that 3.10.x branch possibly local to your environment?
> >
> > I'm not seeing it on my pull...
>
> Ah, yes, never pushed, so should be good to go then on new branch
>
>
> >
> > On Fri, Dec 7, 2018 at 1:52 PM Timothy Bish <ta...@gmail.com> wrote:
> >> On 12/7/18 12:01 PM, Jamie G. wrote:
> >>> Thank you for the additional background information.
> >>>
> >>> So creating a 3.10.x branch off of current 3.9.x branch, than merging
> >>> fixes/features to that branch would be acceptable?
> >> I checked and there already is a 3.10.x branch which I must have been
> >> intending to switch to at some point. You might be best to just drop
> >> that branch and recut from current 3.9.x as that would be the best
> >> starting point for patches. Once moved you might also want to check if
> >> moving the supported APR lib versions up wasn't also a good idea given
> >> the time between releases.
> >>
> >> I think there is some other work you might need to do which is that on
> >> quite a few distros now the cppunit stuff won't build because they've
> >> removed the package configuration script and gone to PKG configs so you
> >> might want to test on more than one distro and or version # (I recall
> >> some reports of problems on Fedora releases after 25 or 26)
> >>
> >> Also if you aren't testing on Windows you should build there and test as
> >> well because that will often show you issues that need addressing before
> >> release. Given the age of the code I'd be surprised if there weren't
> >> some build and test issues that needed addressing.
> >>
> >>> Once a collection of patches have been applied to the 3.10.x, is there
> >>> a release process that out lines how to get this published?
> >>>
> >>> http://activemq.apache.org/cms/creating-distributions.html seems to be
> >>> pre-github.
> >>>
> >>> Cheers,
> >>> Jamie
> >>> On Fri, Dec 7, 2018 at 1:10 PM Timothy Bish <ta...@gmail.com> wrote:
> >>>> My bad, I missed the CC. to dev on this so ignore my comment on keeping
> >>>> in dev. :)
> >>>>
> >>>> On 12/7/18 11:17 AM, Jamie G. wrote:
> >>>>> Hi Tim,
> >>>>>
> >>>>> Can i make master 3.10.0-SNAPSHOT or 4.0.0-SNAPSHOT? Which is preferred?
> >>>>>
> >>>>> In testing code on master the test suites passed on Linux & Mac OSX.
> >>>>> Can you provide more details as to how master is not in a releasable
> >>>>> state? Are there a collection of Jira cards reflecting what needs to
> >>>>> be done to make Master releasable?
> >>>>>
> >>>>>
> >>>>> On Fri, Dec 7, 2018 at 12:29 PM Timothy Bish <ta...@gmail.com> wrote:
> >>>>>> On 12/5/18 6:44 PM, Jamie G. wrote:
> >>>>>>> Hi All,
> >>>>>>>
> >>>>>>> I've started looking into a number of open cards on AMQ CPP client.
> >>>>>>>
> >>>>>>> I've noticed that master is still 3.9.0-snapshot, when it likely
> >>>>>>> should be 3.10.0-SNAPSHOT now.
> >>>>>> Master is not in a releasable state so all work you plan to do to create
> >>>>>> patch releases should be directed to the patch branches unless you plan
> >>>>>> on working on getting master into a releasable state which would be 4.0
> >>>>>> but I doubt you plan to commit that much time to it so I'd suggest
> >>>>>> planning on a 3.x release.
> >>>>>>
> >>>>>>
> >>>>>>> One supplied patch would bump openssl version to 1.0.2 from 0.9.8.
> >>>>>>> Would that bump be sufficient cause to increment the minor version?
> >>>>>>> (https://issues.apache.org/jira/browse/AMQCPP-619)
> >>>>>> 3.9.x is know to work on that older release so publishing a release in
> >>>>>> that series would break the compatibility that would be expected from
> >>>>>> that version range and so it would not be a drop in replacement. The
> >>>>>> more sensible thing to do is to move onto a 3.10.x release series where
> >>>>>> you break that know working range and state that it now requires OpenSSL
> >>>>>> 1.0.x or whatever is the acceptable version. That way a hotfix release
> >>>>>> of 3.9.x is still possible for anyone stuck on old system.
> >>>>>>
> >>>>>>
> >>>>>>> Once these patches are reviewed, and accepted - is there a release
> >>>>>>> process for the AMQ CPP client?
> >>>>>> You would need to update autoconf versions numbers for the release,
> >>>>>> updates code to reflect the release numbers and other things I'm
> >>>>>> probably forgetting. Then you need to test on windows, linux and mac
> >>>>>> with both SSL libs and no SSL libs installed to ensure the changes don't
> >>>>>> break the build and test process. There are both unit tests and
> >>>>>> integration tests that you need to get building and run to validate
> >>>>>> things work on each platform.
> >>>>>>
> >>>>>> Once you validated the code works then you can you the mvn build to have
> >>>>>> a set of archive built which you can then sign and checksum for release.
> >>>>>>
> >>>>>>> Tim, since you did a lot of work on this, your opinion would gladly be
> >>>>>>> appreciated.
> >>>>>>>
> >>>>>>> Cheers,
> >>>>>>> Jamie
> >>>>>>>
> >>>>>> --
> >>>>>> Tim Bish
> >>>>>>
> >>>> --
> >>>> Tim Bish
> >>>>
> >> --
> >> Tim Bish
> >>
>
> --
> Tim Bish
>
Re: [Discuss] ActiveMQ CPP Client
Posted by Timothy Bish <ta...@gmail.com>.
On 12/7/18 12:48 PM, Jamie G. wrote:
> Is that 3.10.x branch possibly local to your environment?
>
> I'm not seeing it on my pull...
Ah, yes, never pushed, so should be good to go then on new branch
>
> On Fri, Dec 7, 2018 at 1:52 PM Timothy Bish <ta...@gmail.com> wrote:
>> On 12/7/18 12:01 PM, Jamie G. wrote:
>>> Thank you for the additional background information.
>>>
>>> So creating a 3.10.x branch off of current 3.9.x branch, than merging
>>> fixes/features to that branch would be acceptable?
>> I checked and there already is a 3.10.x branch which I must have been
>> intending to switch to at some point. You might be best to just drop
>> that branch and recut from current 3.9.x as that would be the best
>> starting point for patches. Once moved you might also want to check if
>> moving the supported APR lib versions up wasn't also a good idea given
>> the time between releases.
>>
>> I think there is some other work you might need to do which is that on
>> quite a few distros now the cppunit stuff won't build because they've
>> removed the package configuration script and gone to PKG configs so you
>> might want to test on more than one distro and or version # (I recall
>> some reports of problems on Fedora releases after 25 or 26)
>>
>> Also if you aren't testing on Windows you should build there and test as
>> well because that will often show you issues that need addressing before
>> release. Given the age of the code I'd be surprised if there weren't
>> some build and test issues that needed addressing.
>>
>>> Once a collection of patches have been applied to the 3.10.x, is there
>>> a release process that out lines how to get this published?
>>>
>>> http://activemq.apache.org/cms/creating-distributions.html seems to be
>>> pre-github.
>>>
>>> Cheers,
>>> Jamie
>>> On Fri, Dec 7, 2018 at 1:10 PM Timothy Bish <ta...@gmail.com> wrote:
>>>> My bad, I missed the CC. to dev on this so ignore my comment on keeping
>>>> in dev. :)
>>>>
>>>> On 12/7/18 11:17 AM, Jamie G. wrote:
>>>>> Hi Tim,
>>>>>
>>>>> Can i make master 3.10.0-SNAPSHOT or 4.0.0-SNAPSHOT? Which is preferred?
>>>>>
>>>>> In testing code on master the test suites passed on Linux & Mac OSX.
>>>>> Can you provide more details as to how master is not in a releasable
>>>>> state? Are there a collection of Jira cards reflecting what needs to
>>>>> be done to make Master releasable?
>>>>>
>>>>>
>>>>> On Fri, Dec 7, 2018 at 12:29 PM Timothy Bish <ta...@gmail.com> wrote:
>>>>>> On 12/5/18 6:44 PM, Jamie G. wrote:
>>>>>>> Hi All,
>>>>>>>
>>>>>>> I've started looking into a number of open cards on AMQ CPP client.
>>>>>>>
>>>>>>> I've noticed that master is still 3.9.0-snapshot, when it likely
>>>>>>> should be 3.10.0-SNAPSHOT now.
>>>>>> Master is not in a releasable state so all work you plan to do to create
>>>>>> patch releases should be directed to the patch branches unless you plan
>>>>>> on working on getting master into a releasable state which would be 4.0
>>>>>> but I doubt you plan to commit that much time to it so I'd suggest
>>>>>> planning on a 3.x release.
>>>>>>
>>>>>>
>>>>>>> One supplied patch would bump openssl version to 1.0.2 from 0.9.8.
>>>>>>> Would that bump be sufficient cause to increment the minor version?
>>>>>>> (https://issues.apache.org/jira/browse/AMQCPP-619)
>>>>>> 3.9.x is know to work on that older release so publishing a release in
>>>>>> that series would break the compatibility that would be expected from
>>>>>> that version range and so it would not be a drop in replacement. The
>>>>>> more sensible thing to do is to move onto a 3.10.x release series where
>>>>>> you break that know working range and state that it now requires OpenSSL
>>>>>> 1.0.x or whatever is the acceptable version. That way a hotfix release
>>>>>> of 3.9.x is still possible for anyone stuck on old system.
>>>>>>
>>>>>>
>>>>>>> Once these patches are reviewed, and accepted - is there a release
>>>>>>> process for the AMQ CPP client?
>>>>>> You would need to update autoconf versions numbers for the release,
>>>>>> updates code to reflect the release numbers and other things I'm
>>>>>> probably forgetting. Then you need to test on windows, linux and mac
>>>>>> with both SSL libs and no SSL libs installed to ensure the changes don't
>>>>>> break the build and test process. There are both unit tests and
>>>>>> integration tests that you need to get building and run to validate
>>>>>> things work on each platform.
>>>>>>
>>>>>> Once you validated the code works then you can you the mvn build to have
>>>>>> a set of archive built which you can then sign and checksum for release.
>>>>>>
>>>>>>> Tim, since you did a lot of work on this, your opinion would gladly be
>>>>>>> appreciated.
>>>>>>>
>>>>>>> Cheers,
>>>>>>> Jamie
>>>>>>>
>>>>>> --
>>>>>> Tim Bish
>>>>>>
>>>> --
>>>> Tim Bish
>>>>
>> --
>> Tim Bish
>>
--
Tim Bish
Re: [Discuss] ActiveMQ CPP Client
Posted by "Jamie G." <ja...@gmail.com>.
Is that 3.10.x branch possibly local to your environment?
I'm not seeing it on my pull...
On Fri, Dec 7, 2018 at 1:52 PM Timothy Bish <ta...@gmail.com> wrote:
>
> On 12/7/18 12:01 PM, Jamie G. wrote:
> > Thank you for the additional background information.
> >
> > So creating a 3.10.x branch off of current 3.9.x branch, than merging
> > fixes/features to that branch would be acceptable?
>
> I checked and there already is a 3.10.x branch which I must have been
> intending to switch to at some point. You might be best to just drop
> that branch and recut from current 3.9.x as that would be the best
> starting point for patches. Once moved you might also want to check if
> moving the supported APR lib versions up wasn't also a good idea given
> the time between releases.
>
> I think there is some other work you might need to do which is that on
> quite a few distros now the cppunit stuff won't build because they've
> removed the package configuration script and gone to PKG configs so you
> might want to test on more than one distro and or version # (I recall
> some reports of problems on Fedora releases after 25 or 26)
>
> Also if you aren't testing on Windows you should build there and test as
> well because that will often show you issues that need addressing before
> release. Given the age of the code I'd be surprised if there weren't
> some build and test issues that needed addressing.
>
> >
> > Once a collection of patches have been applied to the 3.10.x, is there
> > a release process that out lines how to get this published?
> >
> > http://activemq.apache.org/cms/creating-distributions.html seems to be
> > pre-github.
> >
> > Cheers,
> > Jamie
> > On Fri, Dec 7, 2018 at 1:10 PM Timothy Bish <ta...@gmail.com> wrote:
> >> My bad, I missed the CC. to dev on this so ignore my comment on keeping
> >> in dev. :)
> >>
> >> On 12/7/18 11:17 AM, Jamie G. wrote:
> >>> Hi Tim,
> >>>
> >>> Can i make master 3.10.0-SNAPSHOT or 4.0.0-SNAPSHOT? Which is preferred?
> >>>
> >>> In testing code on master the test suites passed on Linux & Mac OSX.
> >>> Can you provide more details as to how master is not in a releasable
> >>> state? Are there a collection of Jira cards reflecting what needs to
> >>> be done to make Master releasable?
> >>>
> >>>
> >>> On Fri, Dec 7, 2018 at 12:29 PM Timothy Bish <ta...@gmail.com> wrote:
> >>>> On 12/5/18 6:44 PM, Jamie G. wrote:
> >>>>> Hi All,
> >>>>>
> >>>>> I've started looking into a number of open cards on AMQ CPP client.
> >>>>>
> >>>>> I've noticed that master is still 3.9.0-snapshot, when it likely
> >>>>> should be 3.10.0-SNAPSHOT now.
> >>>> Master is not in a releasable state so all work you plan to do to create
> >>>> patch releases should be directed to the patch branches unless you plan
> >>>> on working on getting master into a releasable state which would be 4.0
> >>>> but I doubt you plan to commit that much time to it so I'd suggest
> >>>> planning on a 3.x release.
> >>>>
> >>>>
> >>>>> One supplied patch would bump openssl version to 1.0.2 from 0.9.8.
> >>>>> Would that bump be sufficient cause to increment the minor version?
> >>>>> (https://issues.apache.org/jira/browse/AMQCPP-619)
> >>>> 3.9.x is know to work on that older release so publishing a release in
> >>>> that series would break the compatibility that would be expected from
> >>>> that version range and so it would not be a drop in replacement. The
> >>>> more sensible thing to do is to move onto a 3.10.x release series where
> >>>> you break that know working range and state that it now requires OpenSSL
> >>>> 1.0.x or whatever is the acceptable version. That way a hotfix release
> >>>> of 3.9.x is still possible for anyone stuck on old system.
> >>>>
> >>>>
> >>>>> Once these patches are reviewed, and accepted - is there a release
> >>>>> process for the AMQ CPP client?
> >>>> You would need to update autoconf versions numbers for the release,
> >>>> updates code to reflect the release numbers and other things I'm
> >>>> probably forgetting. Then you need to test on windows, linux and mac
> >>>> with both SSL libs and no SSL libs installed to ensure the changes don't
> >>>> break the build and test process. There are both unit tests and
> >>>> integration tests that you need to get building and run to validate
> >>>> things work on each platform.
> >>>>
> >>>> Once you validated the code works then you can you the mvn build to have
> >>>> a set of archive built which you can then sign and checksum for release.
> >>>>
> >>>>> Tim, since you did a lot of work on this, your opinion would gladly be
> >>>>> appreciated.
> >>>>>
> >>>>> Cheers,
> >>>>> Jamie
> >>>>>
> >>>> --
> >>>> Tim Bish
> >>>>
> >> --
> >> Tim Bish
> >>
>
> --
> Tim Bish
>
Re: [Discuss] ActiveMQ CPP Client
Posted by Timothy Bish <ta...@gmail.com>.
On 12/7/18 12:01 PM, Jamie G. wrote:
> Thank you for the additional background information.
>
> So creating a 3.10.x branch off of current 3.9.x branch, than merging
> fixes/features to that branch would be acceptable?
I checked and there already is a 3.10.x branch which I must have been
intending to switch to at some point. You might be best to just drop
that branch and recut from current 3.9.x as that would be the best
starting point for patches. Once moved you might also want to check if
moving the supported APR lib versions up wasn't also a good idea given
the time between releases.
I think there is some other work you might need to do which is that on
quite a few distros now the cppunit stuff won't build because they've
removed the package configuration script and gone to PKG configs so you
might want to test on more than one distro and or version # (I recall
some reports of problems on Fedora releases after 25 or 26)
Also if you aren't testing on Windows you should build there and test as
well because that will often show you issues that need addressing before
release. Given the age of the code I'd be surprised if there weren't
some build and test issues that needed addressing.
>
> Once a collection of patches have been applied to the 3.10.x, is there
> a release process that out lines how to get this published?
>
> http://activemq.apache.org/cms/creating-distributions.html seems to be
> pre-github.
>
> Cheers,
> Jamie
> On Fri, Dec 7, 2018 at 1:10 PM Timothy Bish <ta...@gmail.com> wrote:
>> My bad, I missed the CC. to dev on this so ignore my comment on keeping
>> in dev. :)
>>
>> On 12/7/18 11:17 AM, Jamie G. wrote:
>>> Hi Tim,
>>>
>>> Can i make master 3.10.0-SNAPSHOT or 4.0.0-SNAPSHOT? Which is preferred?
>>>
>>> In testing code on master the test suites passed on Linux & Mac OSX.
>>> Can you provide more details as to how master is not in a releasable
>>> state? Are there a collection of Jira cards reflecting what needs to
>>> be done to make Master releasable?
>>>
>>>
>>> On Fri, Dec 7, 2018 at 12:29 PM Timothy Bish <ta...@gmail.com> wrote:
>>>> On 12/5/18 6:44 PM, Jamie G. wrote:
>>>>> Hi All,
>>>>>
>>>>> I've started looking into a number of open cards on AMQ CPP client.
>>>>>
>>>>> I've noticed that master is still 3.9.0-snapshot, when it likely
>>>>> should be 3.10.0-SNAPSHOT now.
>>>> Master is not in a releasable state so all work you plan to do to create
>>>> patch releases should be directed to the patch branches unless you plan
>>>> on working on getting master into a releasable state which would be 4.0
>>>> but I doubt you plan to commit that much time to it so I'd suggest
>>>> planning on a 3.x release.
>>>>
>>>>
>>>>> One supplied patch would bump openssl version to 1.0.2 from 0.9.8.
>>>>> Would that bump be sufficient cause to increment the minor version?
>>>>> (https://issues.apache.org/jira/browse/AMQCPP-619)
>>>> 3.9.x is know to work on that older release so publishing a release in
>>>> that series would break the compatibility that would be expected from
>>>> that version range and so it would not be a drop in replacement. The
>>>> more sensible thing to do is to move onto a 3.10.x release series where
>>>> you break that know working range and state that it now requires OpenSSL
>>>> 1.0.x or whatever is the acceptable version. That way a hotfix release
>>>> of 3.9.x is still possible for anyone stuck on old system.
>>>>
>>>>
>>>>> Once these patches are reviewed, and accepted - is there a release
>>>>> process for the AMQ CPP client?
>>>> You would need to update autoconf versions numbers for the release,
>>>> updates code to reflect the release numbers and other things I'm
>>>> probably forgetting. Then you need to test on windows, linux and mac
>>>> with both SSL libs and no SSL libs installed to ensure the changes don't
>>>> break the build and test process. There are both unit tests and
>>>> integration tests that you need to get building and run to validate
>>>> things work on each platform.
>>>>
>>>> Once you validated the code works then you can you the mvn build to have
>>>> a set of archive built which you can then sign and checksum for release.
>>>>
>>>>> Tim, since you did a lot of work on this, your opinion would gladly be
>>>>> appreciated.
>>>>>
>>>>> Cheers,
>>>>> Jamie
>>>>>
>>>> --
>>>> Tim Bish
>>>>
>> --
>> Tim Bish
>>
--
Tim Bish
Re: [Discuss] ActiveMQ CPP Client
Posted by "Jamie G." <ja...@gmail.com>.
Thank you for the additional background information.
So creating a 3.10.x branch off of current 3.9.x branch, than merging
fixes/features to that branch would be acceptable?
Once a collection of patches have been applied to the 3.10.x, is there
a release process that out lines how to get this published?
http://activemq.apache.org/cms/creating-distributions.html seems to be
pre-github.
Cheers,
Jamie
On Fri, Dec 7, 2018 at 1:10 PM Timothy Bish <ta...@gmail.com> wrote:
>
> My bad, I missed the CC. to dev on this so ignore my comment on keeping
> in dev. :)
>
> On 12/7/18 11:17 AM, Jamie G. wrote:
> > Hi Tim,
> >
> > Can i make master 3.10.0-SNAPSHOT or 4.0.0-SNAPSHOT? Which is preferred?
> >
> > In testing code on master the test suites passed on Linux & Mac OSX.
> > Can you provide more details as to how master is not in a releasable
> > state? Are there a collection of Jira cards reflecting what needs to
> > be done to make Master releasable?
> >
> >
> > On Fri, Dec 7, 2018 at 12:29 PM Timothy Bish <ta...@gmail.com> wrote:
> >> On 12/5/18 6:44 PM, Jamie G. wrote:
> >>> Hi All,
> >>>
> >>> I've started looking into a number of open cards on AMQ CPP client.
> >>>
> >>> I've noticed that master is still 3.9.0-snapshot, when it likely
> >>> should be 3.10.0-SNAPSHOT now.
> >> Master is not in a releasable state so all work you plan to do to create
> >> patch releases should be directed to the patch branches unless you plan
> >> on working on getting master into a releasable state which would be 4.0
> >> but I doubt you plan to commit that much time to it so I'd suggest
> >> planning on a 3.x release.
> >>
> >>
> >>> One supplied patch would bump openssl version to 1.0.2 from 0.9.8.
> >>> Would that bump be sufficient cause to increment the minor version?
> >>> (https://issues.apache.org/jira/browse/AMQCPP-619)
> >> 3.9.x is know to work on that older release so publishing a release in
> >> that series would break the compatibility that would be expected from
> >> that version range and so it would not be a drop in replacement. The
> >> more sensible thing to do is to move onto a 3.10.x release series where
> >> you break that know working range and state that it now requires OpenSSL
> >> 1.0.x or whatever is the acceptable version. That way a hotfix release
> >> of 3.9.x is still possible for anyone stuck on old system.
> >>
> >>
> >>> Once these patches are reviewed, and accepted - is there a release
> >>> process for the AMQ CPP client?
> >> You would need to update autoconf versions numbers for the release,
> >> updates code to reflect the release numbers and other things I'm
> >> probably forgetting. Then you need to test on windows, linux and mac
> >> with both SSL libs and no SSL libs installed to ensure the changes don't
> >> break the build and test process. There are both unit tests and
> >> integration tests that you need to get building and run to validate
> >> things work on each platform.
> >>
> >> Once you validated the code works then you can you the mvn build to have
> >> a set of archive built which you can then sign and checksum for release.
> >>
> >>> Tim, since you did a lot of work on this, your opinion would gladly be
> >>> appreciated.
> >>>
> >>> Cheers,
> >>> Jamie
> >>>
> >> --
> >> Tim Bish
> >>
>
> --
> Tim Bish
>
Re: [Discuss] ActiveMQ CPP Client
Posted by Timothy Bish <ta...@gmail.com>.
My bad, I missed the CC. to dev on this so ignore my comment on keeping
in dev. :)
On 12/7/18 11:17 AM, Jamie G. wrote:
> Hi Tim,
>
> Can i make master 3.10.0-SNAPSHOT or 4.0.0-SNAPSHOT? Which is preferred?
>
> In testing code on master the test suites passed on Linux & Mac OSX.
> Can you provide more details as to how master is not in a releasable
> state? Are there a collection of Jira cards reflecting what needs to
> be done to make Master releasable?
>
>
> On Fri, Dec 7, 2018 at 12:29 PM Timothy Bish <ta...@gmail.com> wrote:
>> On 12/5/18 6:44 PM, Jamie G. wrote:
>>> Hi All,
>>>
>>> I've started looking into a number of open cards on AMQ CPP client.
>>>
>>> I've noticed that master is still 3.9.0-snapshot, when it likely
>>> should be 3.10.0-SNAPSHOT now.
>> Master is not in a releasable state so all work you plan to do to create
>> patch releases should be directed to the patch branches unless you plan
>> on working on getting master into a releasable state which would be 4.0
>> but I doubt you plan to commit that much time to it so I'd suggest
>> planning on a 3.x release.
>>
>>
>>> One supplied patch would bump openssl version to 1.0.2 from 0.9.8.
>>> Would that bump be sufficient cause to increment the minor version?
>>> (https://issues.apache.org/jira/browse/AMQCPP-619)
>> 3.9.x is know to work on that older release so publishing a release in
>> that series would break the compatibility that would be expected from
>> that version range and so it would not be a drop in replacement. The
>> more sensible thing to do is to move onto a 3.10.x release series where
>> you break that know working range and state that it now requires OpenSSL
>> 1.0.x or whatever is the acceptable version. That way a hotfix release
>> of 3.9.x is still possible for anyone stuck on old system.
>>
>>
>>> Once these patches are reviewed, and accepted - is there a release
>>> process for the AMQ CPP client?
>> You would need to update autoconf versions numbers for the release,
>> updates code to reflect the release numbers and other things I'm
>> probably forgetting. Then you need to test on windows, linux and mac
>> with both SSL libs and no SSL libs installed to ensure the changes don't
>> break the build and test process. There are both unit tests and
>> integration tests that you need to get building and run to validate
>> things work on each platform.
>>
>> Once you validated the code works then you can you the mvn build to have
>> a set of archive built which you can then sign and checksum for release.
>>
>>> Tim, since you did a lot of work on this, your opinion would gladly be
>>> appreciated.
>>>
>>> Cheers,
>>> Jamie
>>>
>> --
>> Tim Bish
>>
--
Tim Bish
Re: [Discuss] ActiveMQ CPP Client
Posted by "Jamie G." <ja...@gmail.com>.
Hi Tim,
Can i make master 3.10.0-SNAPSHOT or 4.0.0-SNAPSHOT? Which is preferred?
In testing code on master the test suites passed on Linux & Mac OSX.
Can you provide more details as to how master is not in a releasable
state? Are there a collection of Jira cards reflecting what needs to
be done to make Master releasable?
On Fri, Dec 7, 2018 at 12:29 PM Timothy Bish <ta...@gmail.com> wrote:
>
> On 12/5/18 6:44 PM, Jamie G. wrote:
> > Hi All,
> >
> > I've started looking into a number of open cards on AMQ CPP client.
> >
> > I've noticed that master is still 3.9.0-snapshot, when it likely
> > should be 3.10.0-SNAPSHOT now.
>
> Master is not in a releasable state so all work you plan to do to create
> patch releases should be directed to the patch branches unless you plan
> on working on getting master into a releasable state which would be 4.0
> but I doubt you plan to commit that much time to it so I'd suggest
> planning on a 3.x release.
>
>
> > One supplied patch would bump openssl version to 1.0.2 from 0.9.8.
> > Would that bump be sufficient cause to increment the minor version?
> > (https://issues.apache.org/jira/browse/AMQCPP-619)
>
> 3.9.x is know to work on that older release so publishing a release in
> that series would break the compatibility that would be expected from
> that version range and so it would not be a drop in replacement. The
> more sensible thing to do is to move onto a 3.10.x release series where
> you break that know working range and state that it now requires OpenSSL
> 1.0.x or whatever is the acceptable version. That way a hotfix release
> of 3.9.x is still possible for anyone stuck on old system.
>
>
> >
> > Once these patches are reviewed, and accepted - is there a release
> > process for the AMQ CPP client?
>
> You would need to update autoconf versions numbers for the release,
> updates code to reflect the release numbers and other things I'm
> probably forgetting. Then you need to test on windows, linux and mac
> with both SSL libs and no SSL libs installed to ensure the changes don't
> break the build and test process. There are both unit tests and
> integration tests that you need to get building and run to validate
> things work on each platform.
>
> Once you validated the code works then you can you the mvn build to have
> a set of archive built which you can then sign and checksum for release.
>
> >
> > Tim, since you did a lot of work on this, your opinion would gladly be
> > appreciated.
> >
> > Cheers,
> > Jamie
> >
>
> --
> Tim Bish
>
Re: [Discuss] ActiveMQ CPP Client
Posted by Timothy Bish <ta...@gmail.com>.
On 12/5/18 6:44 PM, Jamie G. wrote:
> Hi All,
>
> I've started looking into a number of open cards on AMQ CPP client.
>
> I've noticed that master is still 3.9.0-snapshot, when it likely
> should be 3.10.0-SNAPSHOT now.
Master is not in a releasable state so all work you plan to do to create
patch releases should be directed to the patch branches unless you plan
on working on getting master into a releasable state which would be 4.0
but I doubt you plan to commit that much time to it so I'd suggest
planning on a 3.x release.
> One supplied patch would bump openssl version to 1.0.2 from 0.9.8.
> Would that bump be sufficient cause to increment the minor version?
> (https://issues.apache.org/jira/browse/AMQCPP-619)
3.9.x is know to work on that older release so publishing a release in
that series would break the compatibility that would be expected from
that version range and so it would not be a drop in replacement. The
more sensible thing to do is to move onto a 3.10.x release series where
you break that know working range and state that it now requires OpenSSL
1.0.x or whatever is the acceptable version. That way a hotfix release
of 3.9.x is still possible for anyone stuck on old system.
>
> Once these patches are reviewed, and accepted - is there a release
> process for the AMQ CPP client?
You would need to update autoconf versions numbers for the release,
updates code to reflect the release numbers and other things I'm
probably forgetting. Then you need to test on windows, linux and mac
with both SSL libs and no SSL libs installed to ensure the changes don't
break the build and test process. There are both unit tests and
integration tests that you need to get building and run to validate
things work on each platform.
Once you validated the code works then you can you the mvn build to have
a set of archive built which you can then sign and checksum for release.
>
> Tim, since you did a lot of work on this, your opinion would gladly be
> appreciated.
>
> Cheers,
> Jamie
>
--
Tim Bish
Re: [Discuss] ActiveMQ CPP Client
Posted by Roddie Kieley <rk...@gmail.com>.
There is a page at,
http://activemq.apache.org/cms/activemq-cpp-product-version-number.html,
with a few details on the subject, even though it's probably a bit out of
date given that it states it's "DRAFT" status. On the subject of a 'minor'
release it says
"
For the ActiveMQ-CPP project, the numbers have the format
MAJOR.MINOR.REVISION, and the following meanings:
- MAJOR: a change in the MAJOR number is necessary for anything that
changes binary compatibility (e.g. changing a function prototype or class
definition), or when major internal changes occur
- MINOR: a change in the MINOR number is necessary whenever new
functionality is added without breaking binary compatibility (e.g. adding a
new class), or when anything other than very minor internal changes occur
- REVISION: a change in the REVISION number is made for any other type
of change that does not justify a change to MINOR or MAJOR
"
Is there new functionality that comes along with any of the patches or the
OpenSSL bump? There are also some notes regarding the creation of a
distribution for release at
http://activemq.apache.org/cms/creating-distributions.html but again, not
sure how current it would be.
As you mention Tim would probably be the most knowledgeable on the subject.
Roddie
On Wed, Dec 5, 2018 at 8:14 PM Jamie G. <ja...@gmail.com> wrote:
> Hi All,
>
> I've started looking into a number of open cards on AMQ CPP client.
>
> I've noticed that master is still 3.9.0-snapshot, when it likely
> should be 3.10.0-SNAPSHOT now.
>
> One supplied patch would bump openssl version to 1.0.2 from 0.9.8.
> Would that bump be sufficient cause to increment the minor version?
> (https://issues.apache.org/jira/browse/AMQCPP-619)
>
> Once these patches are reviewed, and accepted - is there a release
> process for the AMQ CPP client?
>
> Tim, since you did a lot of work on this, your opinion would gladly be
> appreciated.
>
> Cheers,
> Jamie
>