You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@forrest.apache.org by Thorsten Scherler <th...@apache.org> on 2005/09/03 13:01:36 UTC
Things to enhance as committer
Dear committers,
Disclaimer:
I wrote this mail original to the PMC list, because it is directed to
committers only, and David, as PMC Head, asked me to post it to the dev
list. It is *not* addresses to devs nor user, it it addressed *only* to
*committers*. If you are not a committer I am *not* talking about you.
On Fri, 2005-09-02 at 10:06 +0200, Thorsten Scherler wrote:
> BTW I want to announce that I will lay lower with forrest for a while.
> There are too many things in my real life happening right now that I
> need to leverage my time better. I will still be subscribed but only
> doing work when I can (mostly that will be view related).
On Fri, 2005-09-02 at 18:42 +1000, David Crossley wrote:
> Okay, we can only do what we can. I do hope that
> i did not put you off. That was not my intention.
Actually it is not David, nor any other committer or dev in special. I
will just not have too much spare time in the following months. The time
I have I want to use efficient.
The way we started to handle things in forrest are IMO ATM the opposite
of time efficient.
It seems we prefer to kill a momentum that arises (and that somebody had
a hard time to start) by starting a debate on principles. Latest example
is the lenya forrest integration momentum (which should be the base for
DOCO). What happend to the "Commit-then-review" method for
decision-making?
We say one step at a time, which is perfectly alright, but still we need
to do the first step. If somebody offers to do the first step then try
to lead his way and not building barriers.
It seems we prefer to have endless discussion and well meant
recommendation that do not contain any specific examples (e.g. code
discussion). If you want to change something give an example that is
working and we can discuss this.
If somebody took the time to come up with a solution/recommendation you
should better spend your time to enhance that instead of recommend
something different (e.g. with another focus).
It seems we (committers) open issues that could have been fixed at the
same time as we opening them. Just go ahead and fix it and save the
resources of the ASF and the project (every issue produces overhead:
mail, time, ...).
It seems we (committers) ask whether some components work or not. Just
go ahead and try it (that is what the rest has to do as well).
I would love to keep on helping lenya *and* forrest in the future. I
hope that we can keep the time ratio between discussion and decision as
low as possible. Otherwise it will cost to much time to keep up this
support. Lets please live again our commit-then-review method for
decision-making.
salu2
--
thorsten
"Together we stand, divided we fall!"
Hey you (Pink Floyd)
Re: Things to enhance as committer
Posted by David Crossley <cr...@apache.org>.
Thorsten Scherler wrote:
> Dear committers,
>
> Disclaimer:
> I wrote this mail original to the PMC list, because it is directed to
> committers only, and David, as PMC Head, asked me to post it to the dev
> list. It is *not* addresses to devs nor user, it it addressed *only* to
> *committers*. If you are not a committer I am *not* talking about you.
Great thanks. We want to encourage as much discussion
as possible to happen in the community. The pmc@ private
list is reserved for necessarily private discussions
(such as personal and security issues) and copied when
certain actions happen (such as creating a committer account).
The rest should happen on the dev list, because it gives
the whole community a chance to understand the issues,
thereby making a stronger development community.
That said, if people really do feel uncomfortable
with a certain topic in public, but think something
needs to be addressed, then please do send to pmc@forrest
Otherwise be brave, use the dev list. We are friends.
By the way, Thorsten said "PMC Head". No, that is not
right, it is "chair" i.e. the poor bastard who does
the administrative stuff and trys to keep the meeting
in order and let everyone speak. No-one is head,
we are all equals and each person's opinion has the
same importance.
Here is my reply from that list copied here ...
David Crossley wrote:
> Thorsten Scherler wrote:
> > Dear PMC,
>
> [pmc chair hat on]
> Why is this discussion happening on the PMC mail list?
> I am obliged to ask, every time that it happens. That is
> not the way to do things at ASF. There is nothing private.
> [pmc chair hat off]
>
> The other trouble is that you have a great list
> of recommendations for efficiency below. Unfortunately
> they are now private and the community will not
> know that that is our desire. Hence will eventually
> be repeated out there, hence more mail.
>
> You are absolutely right. All projects need to develop
> ways to be more efficient. Less mail, more doing,
> and leading by example.
>
> This next bit is not directed at anyone in particular.
> I have wanted to say it for a while and it fits here.
> I am concerned that some people might not see the value
> in the non-code discussions. I hope that i am mistaken
> because getting the community practise and composition
> on a correct footing is more important than the code.
> It will lead to efficiency in the long run.
>
> > Thorsten Scherler wrote:
> > > BTW I want to announce that I will lay lower with forrest for a while.
> > > There are too many things in my real life happening right now that I
> > > need to leverage my time better. I will still be subscribed but only
> > > doing work when I can (mostly that will be view related).
> >
> > David Crossley wrote:
> > > Okay, we can only do what we can. I do hope that
> > > i did not put you off. That was not my intention.
> >
> > Actually it is not David, nor any other committer or dev in special. I
> > will just not have too much spare time in the following months. The time
> > I have I want to use efficient.
> >
> > The way we started to handle things in forrest are IMO ATM the opposite
> > of time efficient.
>
> Substitute "We prefer" with "It seems" ...
>
> > We prefer to kill a momentum that arises (and that somebody had a hard
> > time to start) by starting a debate on principles. Latest example is the
> > lenya forrest integration momentum. What happend to the
> > "Commit-then-review" method for decision-making?
>
> That method only relates to code.
>
> I just see discussion in response to some questions
> about which way to go. Sometimes that discussion
> goes longer than it should.
>
> I wonder if it is best to not even ask questions
> about code. Just do it the way one thinks is best.
> Others can change it, or evolve. Of course, there
> are still other times when we do need to discuss.
>
> But maybe you are referring to the vote about
> svn access. The issue is bigger than just giving
> Lenya svn access, so it needed some discussion.
>
> Anyway there is no rush on that, nor reason to
> hold up any development.
>
> > We say one step at a time, which is perfectly alright, but still we need
> > to do the first step. If somebody offers to do the first step then try
> > to lead his way and not building barriers.
>
> I don't see barriers, just constructive discussion.
> If anyone perceives a barrier, then speak up at the
> time please. Nip it in the bud.
>
> > We prefer to have endless discussion and well meant recommendation that
> > do not contain any specific examples (e.g. code discussion). If you want
> > to change something give an example that is working and we can discuss
> > this.
> >
> > If somebody took the time to come up with a solution/recommendation you
> > should better spend your time to enhance that instead of recommend
> > something different (e.g. with another focus).
> >
> > We (committers) open issues that could have been fixed at the same time
> > as we opening them. Just go ahead and fix it and save the resources of
> > the ASF and the project (every issue produces overhead: mail,
> > time, ...).
>
> Yes.
>
> If you are referring to the XHTML2 issues in Jira
> then we decided a while ago that we needed to experiment
> with Jira to assist with co-ordinating our development.
>
> > We (committers) ask whether some components work or not. Just go ahead
> > and try it (that is what the rest has to do as well).
>
> Yes, that is a big one. We need more doing.
>
> > I would love to keep on helping lenya *and* forrest in the future. I
> > hope that we can keep the time ratio between discussion and decision as
> > low as possible. Otherwise it will cost to much time to keep up this
> > support. Lets please live again our commit-then-review method for
> > decision-making.
>
> For most code decisions, yes.
>
> Thanks for taking the time to discuss this.
> Not being facetious, it really is useful.
>
> -David
Re: Things to enhance as committer
Posted by David Crossley <cr...@apache.org>.
Tim Williams wrote:
>
> > It seems we (committers) ask whether some components work or not. Just
> > go ahead and try it (that is what the rest has to do as well).
>
> Call us out. It might be me you're referring to, if I ask something
> obvious, please don't hesistate to call me out on it. I try to read,
> code, then ask as best I can but I'm thick skinned enough to be called
> out on it when I fail to do so.
I think that the suggestion might be regarding
the occasions where we ask whether something works,
rather than just trying for ourselves. For someone
to answer such a question, they would probably need
to try it anyway.
-David
Re: Things to enhance as committer
Posted by Tim Williams <wi...@gmail.com>.
On 9/3/05, Thorsten Scherler <th...@apache.org> wrote:
> Dear committers,
>
> Disclaimer:
> I wrote this mail original to the PMC list, because it is directed to
> committers only, and David, as PMC Head, asked me to post it to the dev
> list. It is *not* addresses to devs nor user, it it addressed *only* to
> *committers*. If you are not a committer I am *not* talking about you.
>
> On Fri, 2005-09-02 at 10:06 +0200, Thorsten Scherler wrote:
> > BTW I want to announce that I will lay lower with forrest for a while.
> > There are too many things in my real life happening right now that I
> > need to leverage my time better. I will still be subscribed but only
> > doing work when I can (mostly that will be view related).
>
> On Fri, 2005-09-02 at 18:42 +1000, David Crossley wrote:
> > Okay, we can only do what we can. I do hope that
> > i did not put you off. That was not my intention.
>
> Actually it is not David, nor any other committer or dev in special. I
> will just not have too much spare time in the following months. The time
> I have I want to use efficient.
>
> The way we started to handle things in forrest are IMO ATM the opposite
> of time efficient.
I think there may be some very valid and valuable reasons for some of
what you perceive to be inefficiencies. I am neither discounting your
points nor saying there's not room for improvement -- my comments that
follow are simply to provide another point of view.
> It seems we prefer to kill a momentum that arises (and that somebody had
> a hard time to start) by starting a debate on principles. Latest example
> is the lenya forrest integration momentum (which should be the base for
> DOCO). What happend to the "Commit-then-review" method for
> decision-making?
When that momentum brings our principles into question, it's not so
bad to debate them. I think there are a couple reasons for this. 1)
to discuss the direction of the momentum itself and 2) to teach new
devs/pmc/committers by openly discussing things (e.g. YOU've taught me
a lot via the latest example you mention and that might not have come
any other way).
It seems to me that our role as PMC members is to look out for the
strategic direction of the project. How can we do that as a whole
without looking at individual actions through the lenses of a
strategic scope and discuss it? Some of us aren't long time members
and just may take a little extra discussion to "know where you're
heading with something" -- a part of the learning process. I
generally don't acquiesce for not knowing -- I prefer to learn and
weigh in based on what's learned.
> We say one step at a time, which is perfectly alright, but still we need
> to do the first step. If somebody offers to do the first step then try
> to lead his way and not building barriers.
i agree -- but more importantly, if you see barriers then highlight it
because I doubt that *anyone* on this project desires barriers, which
leads me to believe that someone is being misunderstood.
> It seems we prefer to have endless discussion and well meant
> recommendation that do not contain any specific examples (e.g. code
> discussion). If you want to change something give an example that is
> working and we can discuss this.
I assume you're talking about view resolver/fallback/lm stuff here. I
think this was covered properly in the other thread. If another dev
has a thought on a way to do something better but doesn't have time to
implement it, would you rather not even hear the idea? As always, you
can simply say "ok, in your spare time, you refactor it that way, but
for now it's working." That, for example, is essentially what Ross
told me when I disagreed with the xslt-based locationmap mounting
stuff IIRC.
> If somebody took the time to come up with a solution/recommendation you
> should better spend your time to enhance that instead of recommend
> something different (e.g. with another focus).
Even if the something different is a better long-term approach?
> It seems we (committers) open issues that could have been fixed at the
> same time as we opening them. Just go ahead and fix it and save the
> resources of the ASF and the project (every issue produces overhead:
> mail, time, ...).
This would be nice right? I mean, who has the time for a *potential*
several hour detour when they're trying to get something done? Some
may seem trivial but often aren't. I long ago learned better than to
say "I'll just fix this bug real quick then I'll get to XYZ.." I
think casting a negative connotation on JIRA Issues is not a good
idea. IMHO experience teaches us to log it -- an additional JIRA
issue doesn't cost that much I think.
> It seems we (committers) ask whether some components work or not. Just
> go ahead and try it (that is what the rest has to do as well).
Call us out. It might be me you're referring to, if I ask something
obvious, please don't hesistate to call me out on it. I try to read,
code, then ask as best I can but I'm thick skinned enough to be called
out on it when I fail to do so.
> I would love to keep on helping lenya *and* forrest in the future. I
> hope that we can keep the time ratio between discussion and decision as
> low as possible. Otherwise it will cost to much time to keep up this
> support. Lets please live again our commit-then-review method for
> decision-making.
I think CTR lives on with respect to code.
As PMC members, our responsibilities to the project and community go
beyond code though. I'd also say that we've taken on quite a few new
folks lately and in doing so, hopefully you all understood that a
along with that would come an increased amount of
questioning/discussion/debate (e.g. email) -- you did think about
that, right? I didn't accept the PMC offer lightly -- I took it
seriously and want to trully understand this stuff so that I can be an
effective PMC member. That, unfortunately, may require some extra
discussion for a while. You guys have had the luxury of working
together for a long time so that you may be able to communicate
meaning in a few words and completely understand each other.
Hopefully I'll get there with everyone too, but for now I prefer a few
extra words/emails to ensure that the complete meaning is being
transmitted. Remember too, that mentoring doesn't end with someone
accepting a PMC offer.
Rather than respond to a whole new email, I'll paste in what David
also said and respond to that too:
> I am concerned that some people might not see the value
> in the non-code discussions. I hope that i am mistaken
> because getting the community practise and composition
> on a correct footing is more important than the code.
> It will lead to efficiency in the long run.
+1, as I look back through the archives, I find some incredibly rich
concept, architecture, design, and implementation discussion happened
that now gives me good insight. I think we owe it to those that end
up picking up our slack to provide the same. Many of the people
active now contributed on those threads that I now lean on. Don't
forget where <generally>you<generally> started -- lot's of googling
with "site:marc.theaimsgroup.com" as the parameter, right?
On a side note, your email contains a lot of apparent references to
real things. I personally would prefer that you just correct me
straight out when I do something dumb (e.g. ask an obvious question)
instead of letting it build up. I've got very thick skin and won't be
offended with a "RTFM"-type response when it's appropriate.
--tim
Re: Things to enhance as committer
Posted by Ross Gardler <rg...@apache.org>.
Thorsten Scherler wrote:
> It seems we prefer to kill a momentum that arises (and that somebody had
> a hard time to start) by starting a debate on principles. Latest example
> is the lenya forrest integration momentum (which should be the base for
> DOCO). What happend to the "Commit-then-review" method for
> decision-making?
I have a great deal of sympathy with this. I also recognise there should
be a balance between making major contributions and ensuring those
contributions are well thought through. This is a difficult thing to
balance. We do get it wrong sometimes.
My response here is not intended as an argument, but an attempt to add a
little balance to this particular position, as I said, I have much
sympathy with Thorstens intentions with this mail and add my support to
that intention.
> We say one step at a time, which is perfectly alright, but still we need
> to do the first step. If somebody offers to do the first step then try
> to lead his way and not building barriers.
Building barriers is certainly the *wrong* balance.
> It seems we (committers) open issues that could have been fixed at the
> same time as we opening them. Just go ahead and fix it and save the
> resources of the ASF and the project (every issue produces overhead:
> mail, time, ...).
Speaking personally I know I open issues that I see whilst working on
other parts of code. When I'm working on one job I cannot start fixing
things that are not directly related to that job. Creating an issue is a
way of logging the issue so it is not forgotten.
I suspect the same is true for everyone. To fix an issue requires
testing (see the bug I introduced in views with my *.source.xml patch
that I did for a specific job but clearly did not test enough).
> Lets please live again our commit-then-review method for
> decision-making.
(now I am speaking personally...)
If I'm not sure of the best way to implement something I will always ask
questions first. For major changes I will *always* ask questions first,
I don't like doing something twice and I want to ensure my contributions
are the right thing to do.
For smaller contributions, yes, I agree, commit-then-review.
Ross