james-server-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Niklas Therning <nik...@trillian.se>
Subject Re: [mime4j] what else need to be done (Was: Benchmark of 0.3 vs 0.4)
Date Wed, 06 Aug 2008 08:39:02 GMT
Stefano Bagnara wrote:
> Robert Burrell Donkin ha scritto:
>> On Tue, Aug 5, 2008 at 6:16 PM, Bernd Fondermann
>> <bernd.fondermann@googlemail.com> wrote:
>>> On Tue, Aug 5, 2008 at 18:52, Stefano Bagnara <apache@bago.org> wrote:
>> <snip>
>>>> IMHO this one (use the new structure but move some class from 
>>>> parser to
>>>> main) is the worst of the 5 analyzed, but I won't veto it, so if 
>>>> this is
>>>> what the majority wants, I'm fine with it.
>>> I don't think that Apache is about majority decisions in the first
>>> place. Majority decisions (votes) are often not including minority
>>> opinions, while finding consensus is about taking every opinion into
>>> acount (known as 'compromising'). If that doesn't work, votes come
>>> into play. (Disclaimer: The readers notion about 'The Apache Way' may
>>> vary.)
>> i think it's important to let people know about changes like this and
>> assemble some kind of rough consensus before embarking. i think
>> there's now a consensus that these changes are broadly an improvement
>> but that more refinements are possible. let's just apply the proposal
>> and then start working on the improvements.
> MIME4J-51 proposal has already been applied, unfortunately what I 
> thought was consensus a week ago now seems something invented by me ;-)
> My questions to Niklas and Bernd are really to understand if 
> "refinements" is something that can be done on the refactored tree or 
> if the consensus is now too far from that and we should better revert 
> it and start from that.
> I'm scared that people think I'm pushing my proposal or pushing my 
> cycle dependencies issues when I just want to see something done.
> I hope Niklas will update us soon, without fears or bad feelings wrt 
> this thread, and I also hope that someone else with better 
> communications skills can show what's wrong in the way I try to build 
> consensus.


I certainly understand the merits of trying to reduce cyclic package 
dependencies, especially for the developers developing the software. My 
only concern is that zero cyclic dependencies in a library such as 
Mime4j doesn't necessarily mean that it will be easier to use for the 
end user.

I definitely think that most of the refactorings you've made are great. 
My suggestion was only to move the four classes/interfaces I mentioned 
to the root package. That's all. I've played around with JDepend and if 
I understand correctly this change results in a single cycle (mime4j <-> 
parser). My gut feeling is that this would make it easier for the end 
user. However, as an active committer I'm not the typical end user 
myself so my gut feeling could be wrong. I can definitely live with what 
is in trunk right now. To me it is more important that we release a new 
version soon. Also, at this stage of the project I think we can afford 
to (and probably we will) change our minds about the package hierarchy a 
couple of times more as the code evolves.

I don't think there is anything wrong with your communication skills. 
I'm impressed that you are investing so much of your personal time in 
trying to make Mime4j a better piece of software. Don't stop doing that! 
:) The reason why I didn't raise my concerns earlier is that I haven't 
had the time to catch up until now (I just come back from vacation).


To unsubscribe, e-mail: server-dev-unsubscribe@james.apache.org
For additional commands, e-mail: server-dev-help@james.apache.org

View raw message