If this email is too long for you but you want to read at least a part
of it, please skip to the last part (bottom line is...).


----- Original Message -----
From: "Noel J. Bergman" <[EMAIL PROTECTED]>


> > No the intent was not to attack, but to make sure the developer
> > focusses on testing what he commits.
>
> That is not how others perceived it.  If your intent is to simply report a
> defect, then we need to have a talk about inter-personal communication.
> Perhaps we need to talk about how to, and how not to, report defects
similar
> to the Eric Raymond paper on how to ask questions.

Noel,

Here is what I see,

- There have been interpersonal issues in James in the last 3 months
  or so. Before that there has not been a single interpersonal issue
  in the last 2 years of my involvement with James.
- All the interpersonal issues have one common developer.

I think that particular developer should make an attempt to work with
others and try to get the best out of them. Different committers seem
to have exposed what appears to be a negetive side, and that is a new
thing. I don't think this happened because a few year old James
developers decided to loose politeness in Q4 2002. This is not a
negetive attack but a suggestion that is likely to be more beneficial
than any process that may come out of Eric's paper or Sam's email.

Attempt should be made to play to strengths of individual developers,
tap into their knowledge and move the project forward without
complaing about lack of participation. Do your best and assume others
are doing their best too. If someone can't do better they probably
have personal or other issues to deal with.



>
> > If it works all is good.
>
> No, all is not good.  Aaron isn't the only person who is offended enough
to
> consider leaving the project because of the bickering.
> THAT is the damage,

True. Others have been offended by this too and have considered
leaving. I know at least one more person. This lack of responsibility
and bickering sucks energy and must cease.


> not some minor defect in protocol handling.

Breaking a protocol to the point of making it useless repeatedly
without any testing is not a minor defect.


>
> > This is the 3rd issue that completly breaks NNTP Server that I have
found
> so far.
> > Testing is expected behaviour ...
>
> And was expected in October.  According to you, this is such a obvious
issue
> that any client test in that time frame should have shown it.  Apparently,
> our users don't use NNTP, otherwise didn't encounter this issue, or didn't
> care enough about NNTP to report it.

Since 2.1 announcement there have been 2 users of NNTP on the Users
list. Both had (some)success with earlier version. That is what got my
attention and an attempt to test NNTP. The earliest I could get to it
was the holiday season.

I can think of at least 2 other reasons for this.
- NNTP is marked as experimental in the docs. NNTP is not
  experimental. It has been a working part of James for some
  time. There have been users. Docs should have changed but was not
  done and this I believe discourages users.
- NNTP is not as important as SMTP/POP3

>
> Actually, there are other less critical flaws in NNTP, one of which is in
> Bugzilla.  I believe that fixes exist and are pending the v2.1 release.
So
> the quicker we ship this update, the quicker we can get to other issues.

The bug fix you sent is the reverse of this change:
http://cvs.apache.org/viewcvs.cgi/jakarta-james/src/java/org/apache/james/nn
tpserver/repository/NNTPArticleImpl.java.diff?r1=1.9&r2=1.10
This checkin broke protocol and hence news readers. The checkin comments are
misleading at best.
I hope the fix has better quality. I would like to review it if possible.




The bottom line is this:

Testing has revealed an issue and instead of addressing we seem to be
going in different directions. Instead of disabling NNTP or shipping a
broken version, I want to fix this. It may involve rolling back some
code but I will try to keep as much as possible. Again will be happy
if someone else wants to fix it.


I have still not heard if the person who has broken NNTP if he intends
to fix it. My tests tell me the bug fix reverses a regression that was
caused
by the above mentioned checkin but does not fix the bug (. I wish the
(ir)responsible developer would fix the bug rather than throw
something over the fence via an indirect channel without testing(again).
Don't enjoy testing patches that promise to fix the bug but don't seem
to have been tested.

This is not fair. It is ~ 3:00 on a Friday night. I am trying to test
stuff that should not have broken in the first place. I could have in
less time fixed the server. I am testing, explaining mostly to get
everybody on board and give the unnamed contributor a chance to fix
this. This explaining actions to death is a direct result of the
friction in James over the last few months. I don't think I would have
any need to explain as much earlier, would have just fixed it.

Harmeet


--
To unsubscribe, e-mail:   <mailto:[EMAIL PROTECTED]>
For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>

Reply via email to