I get the sense that both of you are a bit frustrated. The transition to the new RTC development
model has been challenging for all. The PMC has not kept up with the number of reviews and that has
allowed the codebase to drift while patches then get stale.
Recently we've made several steps forward to improve the process. Several new PMC members have been
added in the last few weeks that are active committers to the project so the ability to provide
timely feedback has been improved. Along with some better mechanisms to track what needs to be
reviewed so we can quickly address them. (I think we got the pluggable JAAC completed last night).
All this to say it has made us stumble a bit in the process, we're not perfect yet, but we're making
some important headway.
Everyone who has worked on the Maven 2 conversion needs to get some kudos as it is slightly larger
than a bread box :) and given the somewhat binary nature of the change does not lend itself well to
using patches as the vehicle to get the job done.
All that said, Jason, as your going through the patches and making changes what is the primary way
to get feedback to the person who contributed the patch? It sounds like you have some good feedback
that would help Anita produce patches that you are both in more agreement on. Also, it sounds like
some of the changes are preferences based on style. It would be a fair debate as to one should use
Plexus or Geronimo infrastructure for the file delete activity.
All this said, you guys are to be commended for the progress you've made. For the time being the
review and collaboration feels like we've gone from a sprint to a jog but as we hit our stride I
hope the pace will pick up as well.
anita kulshreshtha wrote:
inline..
--- Jason Dillon <[EMAIL PROTECTED]> wrote:
On Aug 7, 2006, at 10:33 PM, anita kulshreshtha wrote:
This code is from servlets-examples-jetty config (rev 429124):
<resources>
<resource>
<directory>${pom.basedir}/src/conf</directory>
<targetPath>META-INF</targetPath>
<includes>
<include>geronimo-plugin.xml</include>
</includes>
<filtering>true</filtering>
</resource>
</resources>
This code has been added to many applications config. Which
means
that you are trying to write it yourself and have no intention of
using
the patch.
I was simply reusing the existing Maven2 resources plugin to handle
filtering of resources.
This high quality code does not do what it is supposed to do, i.e. put
geronimo-plugin.xml in the generated car.
I looked over your patch and could not apply it directly due to the
number of other changes made to the tree since the patch was
originally crafted.
Why did you ask me to make the patch?
I asked you to roll new patches against m2migration and not off of
trunk so that I could quickly verify and apply them.
The patch on July 27th was for m2migration and it is clearly written.
Wow.. I don't blame you for exercising the power of a committer.
you
get to commit code that does nothing and reject the code that
works!
You have the power to shut down other peoples work.
I am starting to take offense to some of these comments you are
making. I'm not sure if you are trying to goat me into a conflict or
if you are trying to resolve the work you have done and move forward.
:-(
Jason, I was also aware of the issues with the code and had been
wanting to fix them and add more functionality. You are constantly
changing the code that I wrote without any communication. You have
made
it _impossible_ for me to work on this code. I am not saying that
you
are doing it intentionally.
Since these commits end up with my user id attached to them, I am not
willing to commit something that does not meet my standards for
quality. I am not trying to invalidate your work, I am trying to get
our m2 build functional and at the same time ensure a high standard
of quality for the code that supports it.
FYI, the code you are talking about was already committed by David
Jencks! David helped me write the plugin by expaining how the configs
work. He patiently reviewed massive patches, tested them, committed
them and made sure that the first server could be started.
IMO, you should have accepted the code
because it provided the required functionality and allowed me to
make
improvements.
The code submitted in the patches that I reviewed (and some that I
committed and then changed) were not using the Mojo API appropriately
or effectively. Just because a chunk of code "works" does not mean
that it should be blindly applied to the tree.
Isn't it because you added Mojo for the code that is not even being
used?
I accepted the bulk of the code and cleaned it up to meet my
standards before I committed it. Though some of your code I have not
even begun to review since it is scattered amongst several issues and
then into several patches in those issues, which makes it much harder
for me to quickly verify and commit.
Last time I checked the new patches are still using velocity and
custom file deletion bits instead of using the existing Plexus
support tools that handle this for you....
The file deletion code is straight out of geronimo! If it is being used
in Geronimo, it is good enough for me. Why would I use Plexus?
BTW, many months ago *when you were not involved in this effort*, David
suggested that we should generate classpath dynamically. I was waiting
for the first server to start before I attempted that. That is why the
classPath was used as a massive string. It was only temporary.
and nothing is commented.
So it is much more difficult for me to simply commit this.
I agree with Hiram Chirino on this subject. I am quoting
from a conversation on the list :
http://www.nabble.com/Re%3A--RTC--ActiveMQ-GBean-modules-p4867711.html
"Perhaps I should start a new thread on this thought, but I just
wanted
to comment that we need to be careful about how critical and the
level
of perfection that we expect from the contributed patches. I would
say that if a patch does not regress the project and it moves it
forward in the right direction, the patch should be accepted even
if
it's not perfect.
It kind of reminds me of something David B told me once, if the
code
is perfect and stable, you won't be able to build a community
around
the project it since it just works. This makes sense to me. If
the
code is 80% of the way there, then you give an opportunity for
folks
to join your community by submitting additional patches that help
it
get to the 100% mark."
I generally agree with Hiram, though I don't think that we can allow
build infrastructure related patches of diminished quality to be
applied with out retrofitting them... or we will just make a larger
mess for everyone to deal with.
* * *
I am sorry that you are upset about the situation related to your
patches. I would really like for us to get past this and get back to
being productive.
But, to be honest with you... the more defensive emails like this
that you post, the less I want to continue working on the related
issues. I want to get the m2 work behind us and not get bogged down
with conflict with those who are helping that work.
Again, I am sorry you are upset... but can we please try to move
forward?
Unfortunately we will have to revisit this until all the configs have
the required functionality! You see, When they don't work people file
issues. I am waiting patiently...
Thanks
Anita
--jason
__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com
