karaf-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Christian Schneider <ch...@die-schneider.net>
Subject Re: bnd files in Decanter Project
Date Thu, 11 Feb 2016 19:19:07 GMT
Hi Achim,

thanks for your detailed description. I can assure you that I do not 
follow any evil plot like you seem to think. I will for sure not use 
gradle any time soon. I am quite happy with maven and am just trying to 
make our project better and the code more efficient. I also was not 
really sneaky by creating an issue up front and describing what I want 
to do. I just did not to also spawn a discussion on the list to make 
this more obvious. Please be assured that the only reason for this was 
my inherent impatience and good faith that JB would not have supported 
my change if he thought it would be a bad thing for the karaf community. 
So I was not really aware of doing something wrong at this time.

Please also take into account that your build problem only happened on a 
new module you wanted to add and could be fixed by just adding the empty 
bnd.bnd file. So I think a -1 while possible was not really necessary. I 
think the better option would have been to bring the issue on the list 
and try to reach a consensus. I think in my whole apache history I never 
did a -1 on a commit.

Do you insist on me taking back the commit now or would it be ok to 
first try to reach a consensus or if not possible do a vote? I will 
happily undo my commit if the outcome is to not do the change. On the 
other hand if we agree to do the change it would save me to shuffle the 
code back and  forth.

If you insist on this I will undo the changes now. It is probably not 
possible to just undo the commit as there are quite a few commits in 
between. So it will be quite some manual work for me.
I also hope that if you insist on me taking back the change now you are 
ok if I only undo the bnd file part of the commit and leave other 
changes like moving some common deps into the parent in place.


On 11.02.2016 17:06, Achim Nierbeck wrote:
> Hi there,
> I'll stick to my
> -1
> Now the reasons for this:
> Even though the opposite has been proclaimed on the list, it did break the
> build for me. As without a bnd file the project isn't building
> successfully, even though the information needed has been provided by the
> pom.
> Second the usage of an extra bnd file is error prone, and I don't care
> about removed lines of code, this is just not a justification.
> Another thing why I opposed to extra bnd files. Only on Eclipse when using
> the bnd-tools project you have an extra editor available to use it. And
> this only works for people using both, eclipse and the bnd-tools. All
> others using a different tool chain are losing the benefit to see everthing
> in one go.
> So basically people using IntelliJ or Netbeans (both tools to be know to
> work far better with maven, then eclipse) are out of this. And I rather
> don't rely on a proprietary tooling for this.
> So what's next after this? Just sneak the bnd-tools plugins into this
> project for "supposedly" better support with eclipse and bnd-tools?
> I don't think so. Our basis is still Maven and not Gradle, and this is
> because we have a big user-base that feels very comfortable with it.
> Using the declarations in the pom has been working nicely and if nothing
> needs to be done one just needs to add those 4 lines for the usage of the
> maven-bundle-plugin. No extra file needed, everything is in one place and
> is in line with all other Karaf projects.
> Actually this is one of the reasons I removed those bnd files from the
> pax-web project long time ago, because all those project related
> information are in one place now, instead of scattered around the project.
> Third, and this does outweigh everything else. Christian tried to provide
> evidence with this move and now argues this way everything is better for
> us.
> At this point I can't accept this. I don't think this is the way our
> community works and surely this isn't the way we work together on our
> sources. At least
> this has been my understanding of a successful project.
> If I'm wrong on this and everybody can do as he likes, I guess I need think
> about the consequences.
> regards, Achim
> 2016-02-11 15:12 GMT+01:00 Jean-Baptiste Onofré <jb@nanthrax.net>:
>> I see valid arguments here, and I keep my +1.
>> Regards
>> JB
>> On 02/11/2016 02:40 PM, Christian Schneider wrote:
>>> I also did a jira issue that explains the change.
>>> https://issues.apache.org/jira/browse/KARAF-4300
>>> To show the advantage let me compare one of the old configs to the new
>>> one:
>>> http://apaste.info/gC5
>>> This shows that 18 lines of xml are replaced by 2 lines in the bnd file.
>>> The syntax is basically the same
>>> as in xml just without the brackets. The big advantage is that we can
>>> leave out the boilerplate xml that maven
>>> needs to redefine a plugin.
>>> So I think the change to bnd files makes a lot of sense. I also did this
>>> change in Aries a long time ago.
>>> Btw. I also added API baselining in this change which helps us when do
>>> changes on the API.
>>> Christian
>>> On 11.02.2016 14:21, Christian Schneider wrote:
>>>> As further reference. This is the commit where I did the change:
>>>> https://github.com/apache/karaf-decanter/commit/dabeeecd41f46a0c5344580f65c2ea6877fd6d35
>>>> As you can see I added 108 lines and removed 753.
>>>> That means the usage of bnd files saves us about 640 lines of xml. I
>>>> think this is a strong indicator that it is a good idea to do so. Of
>>>> course it might cause problems that I overlooked.
>>>> Christian
>>>> On 11.02.2016 13:57, Christian Schneider wrote:
>>>>> Hi Achim,
>>>>> it is difficult to predict what changes warrant a discussion. I agree
>>>>> with you that I should have discussed this on the list. I talked to
>>>>> JB and he was positive so I did not expect any problems. Apparently I
>>>>> was wrong :-(
>>>>> The usage of bnd files for configuration of the imports and exports
>>>>> is a very concise replacement for the same configs in xml.
>>>>> The big advantage is that you can omit the maven-bundle-plugin in the
>>>>> pom of each module. So basically you replace about 10-15 lines of xml
>>>>> with 0-5 lines in the bnd file.
>>>>> This does not break any functionality for users. For developers it
>>>>> just requires to put an empty bnd.bnd file into each module if it
>>>>> does not need special configs. Unfortunately it is not possible to
>>>>> define that it uses a bnd file if it is there  and is also happy if
>>>>> no such file is there. I plan to suggest this to the felix project as
>>>>> it would make this even easier to use.
>>>>> In what way do you see this as a breaking change? I made sure that
>>>>> all code still works and that all tests still pass. I also did manual
>>>>> tests of all the modules.
>>>>> So the only thing you need to do for a new module is to add this
>>>>> empty bnd.bnd file and you are fine. If you do not want to use the
>>>>> bnd file to configure your OSGi configs you can use the xml like before.
>>>>> So apart from my bad communication where I fully agree with you.. do
>>>>> you really  think this warrants a -1?
>>>>> Do you have any technical concerns?
>>>>> Christian
>>>>> 2016-02-11 9:55 GMT+01:00 Achim Nierbeck <bcanhome@googlemail.com
>>>>> <mailto:bcanhome@googlemail.com>>:
>>>>>      Hi,
>>>>>      the other day I added another module to the decanter project
>>>>>      (cassandra
>>>>>      appender).
>>>>>      And I've got to say I was quite astonished to see all those bnd
>>>>>      files in
>>>>>      there, but what
>>>>>      really got me stirred. It is mandatory to have those now.
>>>>>      I can't remember seeing a vote for such a change in development!
>>>>>      So here is my
>>>>>      -1
>>>>>      on this not communicated and breaking functionality change that
>>>>>      sneaked in
>>>>>      there.
>>>>>      So whoever changed that needs to revoke this, NOW.
>>>>>      It hasn't been discussed up-front and actually I just can't stand
>>>>>      such
>>>>>      sneaky moves.
>>>>>      regards, Achim
>>>>>      --
>>>>>      Apache Member
>>>>>      Apache Karaf <http://karaf.apache.org/> Committer & PMC
>>>>>      OPS4J Pax Web <http://wiki.ops4j.org/display/paxweb/Pax+Web/>
>>>>>      Committer &
>>>>>      Project Lead
>>>>>      blog <http://notizblog.nierbeck.de/>
>>>>>      Co-Author of Apache Karaf Cookbook <http://bit.ly/1ps9rkS>
>>>>>      Software Architect / Project Manager / Scrum Master
>>>>> --
>>>>> --
>>>>> Christian Schneider
>>>>> http://www.liquid-reality.de
>>>>> <
>>>>> https://owa.talend.com/owa/redir.aspx?C=3aa4083e0c744ae1ba52bd062c5a7e46&URL=http%3a%2f%2fwww.liquid-reality.de
>>>>> Open Source Architect
>>>>> http://www.talend.com
>>>>> <
>>>>> https://owa.talend.com/owa/redir.aspx?C=3aa4083e0c744ae1ba52bd062c5a7e46&URL=http%3a%2f%2fwww.talend.com
>>>> --
>>>> Christian Schneider
>>>> http://www.liquid-reality.de
>>>> Open Source Architect
>>>> http://www.talend.com
>> --
>> Jean-Baptiste Onofré
>> jbonofre@apache.org
>> http://blog.nanthrax.net
>> Talend - http://www.talend.com

Christian Schneider

Open Source Architect

View raw message