[U-Boot-Users] CVS Write Access to Das U-Boot
Dear Wolfgang,
As the maintainer of the ARMboot project I welcome your effort to merge the PPCboot and ARMboot sources and making them available within the U-Boot project. Here at SYSGO, we are ready to discontinue the ARMboot project in favour of U-Boot, if we can continue our work on the U-Boot source tree.
Therefore I would appreciate to have CVS write access to the U-Boot's source repository on SF.NET. Efficient Open Source development requires working along the "Release Early, Release Often" scheme. This is severely hampered if all updates are channelled through one person, who might not always be available.
With the software already growing faster than we all ever thought possible, it only makes sense to split responsibilities. For that reason, we would like to keep our attention on the ARM portions on the tree. Obviously, changes in the common code must be discussed first with the maintainers, better yet, with the community.
Also, to be honest, it appears a bit rude to me to take our ARMboot sources, and then just cut us off. It seems only natural to me that we both continue doing what we have done all along, only with more benefit for all of us.
Please let's keep this an _open_ project, not one where only one member has the power.
Best Regards, Marius
----------------------------------------------------------------------------- Marius Groeger SYSGO Real-Time Solutions AG mgroeger@sysgo.de Software Engineering Embedded and Real-Time Software www.sysgo.de Voice: +49-6136-9948-0 Am Pfaffenstein 14 www.osek.de FAX: +49-6136-9948-10 55270 Klein-Winternheim, Germany www.elinos.com
Dear Marius,
in message Pine.LNX.4.40.0211041505080.2379-100000@mag.devdep.sysgo.de you wrote:
As the maintainer of the ARMboot project I welcome your effort to merge the PPCboot and ARMboot sources and making them available within the U-Boot project. Here at SYSGO, we are ready to discontinue the ARMboot project in favour of U-Boot, if we can continue our work on the U-Boot source tree.
You are welcome.
Therefore I would appreciate to have CVS write access to the U-Boot's source repository on SF.NET. Efficient Open Source development requires working along the "Release Early, Release Often" scheme. This is severely hampered if all updates are channelled through one person, who might not always be available.
Your argument is valid, but as I already explained to Robert Kaiser, it is not the only valid argument; here is a famous quote:
Conceptual integrity in turn dictates that the design must proceed from one mind, or from a very small number of agreeing resonant minds. - Frederick Brooks Jr., "The Mythical Man Month"
Before I add anybody to the project as developer with write permissions to CVS, I want to make sure that there is at least a minimul level of such "resonance". This includes simple things like conforming to a common coding style, in-advance negotiation of changes with global impact, etc.
From my experience with PPCBoot I can tell you that there are many
talented programmers, but I definitely would not want to give CVS write permissions to more maybe 10% of them.
That does NOT mean that there is any discrimination, or that any con- tributions will be igrnored or turned down - if you want, ask if any of the people who contributed to PPCBoot has reason to complain.
Please start for now by sending patches like anybody else. Normally there will be no significant delay introduced by my moderation. You can be pretty sure that I'll want to add you as developer rather sooner or later. But I think fairness requires to apply the same rules to everybody.
Also, to be honest, it appears a bit rude to me to take our ARMboot sources, and then just cut us off. It seems only natural to me that we
Umm, calm down. On the same base I could argument that you cut us (the PPCBoot guys off) by taking "our" PPCBoot sources and creating an incompatible ARMBoot. But I don't intend to start such a fruitless discussion.
Let's focus on what what we'll have to do in the future.
both continue doing what we have done all along, only with more benefit for all of us.
Definitely.
Please let's keep this an _open_ project, not one where only one member has the power.
Umm... please don't misinterpret my role as a moderator.
It does NOT require write permission in the CVS to contribute to a project, or to get credited for your work.
Best regards,
Wolfgang Denk
On Mon, 4 Nov 2002, Wolfgang Denk wrote:
Before I add anybody to the project as developer with write permissions to CVS, I want to make sure that there is at least a minimul level of such "resonance". This includes simple things like conforming to a common coding style, in-advance negotiation of changes with global impact, etc.
I thought we have known each other for long enough a time to have such "trust" in each other's work. We will probably never gain 100% resonance, but then I wouldn't expect this anyway. The success of team-work is determined by the ability to make some compromises here and there. To stick with the picture: the resonnance frequency is somewhere between the developers.
That does NOT mean that there is any discrimination, or that any con- tributions will be igrnored or turned down - if you want, ask if any of the people who contributed to PPCBoot has reason to complain.
That's none of my business. All I'm saying is that having worked together for a long time, and us having ported ARMboot in the first place should suffice for us to have write access. And somehow I believe you know I'm right.
Umm, calm down. On the same base I could argument that you cut us (the PPCBoot guys off) by taking "our" PPCBoot sources and creating an incompatible ARMBoot. But I don't intend to start such a fruitless discussion.
I aggree. The mailing lists dcoument well enough why things went like they did. You never asked me for write access, though, that I would have granted you immediately.
Please let's keep this an _open_ project, not one where only one member has the power.
Umm... please don't misinterpret my role as a moderator.
It does NOT require write permission in the CVS to contribute to a project, or to get credited for your work.
Really? I remember some lengthy discussions about a patch that ended with a rather, er, grungy moderator asking me multiple times for a rewording of the README part of the patch, because he didn't like the patch.
There _is_ a difference if you have to beg for a feature to go into the sources.
Best Regards, Marius
----------------------------------------------------------------------------- Marius Groeger SYSGO Real-Time Solutions AG mgroeger@sysgo.de Software Engineering Embedded and Real-Time Software www.sysgo.de Voice: +49-6136-9948-0 Am Pfaffenstein 14 www.osek.de FAX: +49-6136-9948-10 55270 Klein-Winternheim, Germany www.elinos.com
In message Pine.LNX.4.40.0211041843090.2621-100000@mag.devdep.sysgo.de you wrote:
I thought we have known each other for long enough a time to have such "trust" in each other's work. We will probably never gain 100%
Let's work a while togehter on this project, and we'll see.
Really? I remember some lengthy discussions about a patch that ended with a rather, er, grungy moderator asking me multiple times for a rewording of the README part of the patch, because he didn't like the patch.
So probably it's not exactly a "resonant mind" then.
There _is_ a difference if you have to beg for a feature to go into the sources.
I rather discuss such an issue before, and then have an agreement of the involved parties. This has nothing to do with "begging".
You mention exactly one of the reason why I don't intend to grant you or anybody else immediate CVS write access. I want to know how it works out exactly in situations like that.
What's so bad with sending a couple of patches first? Everybody else does.
Let's stop this discussing, it costs but time. If you want to help, please send patches, and you are welcome.
Best regards,
Wolfgang Denk
On Mon, 4 Nov 2002, Wolfgang Denk wrote:
Really? I remember some lengthy discussions about a patch that ended with a rather, er, grungy moderator asking me multiple times for a rewording of the README part of the patch, because he didn't like the patch.
So probably it's not exactly a "resonant mind" then.
So you define the resonance frequency. Which I find not very cooperative. Don't we have the same rights, like to reject (or debate) _your_ enhancements?
What's so bad with sending a couple of patches first? Everybody else does.
Let me ask you a question: why do you maintain your own kernel tree? Probably because its easier than getting your patches accepted. I want to avoid that situation happening again in our micro-cosmos. (And that's what it really is, be fair.)
Let's stop this discussing, it costs but time. If you want to help,
Wolfgang, please don't always turn away from inconvenience like that. I know it's tedious, and I know that some people out there shake their heads, but we're dealing with Open Source here, so the rights of maintainance should be discussed openly as well. Actually I'd be interested in other peoples opinion on this. Do we need aprobation time to be allowed to continue working on U-Boot[ARM] again?
Best Regards, Marius
----------------------------------------------------------------------------- Marius Groeger SYSGO Real-Time Solutions AG mgroeger@sysgo.de Software Engineering Embedded and Real-Time Software www.sysgo.de Voice: +49-6136-9948-0 Am Pfaffenstein 14 www.osek.de FAX: +49-6136-9948-10 55270 Klein-Winternheim, Germany www.elinos.com
In message Pine.LNX.4.40.0211041930120.2979-100000@mag.devdep.sysgo.de you wrote:
What's so bad with sending a couple of patches first? Everybody else does.
Let me ask you a question: why do you maintain your own kernel tree? Probably because its easier than getting your patches accepted. I want
You know the reasons very well:
We need a stable kernel for our customers, but patches and extensions get only accepted for the 2.5 kernel, which is far from acceptable for a customer project. So we simply have no choice but maintaining our own trees.
interested in other peoples opinion on this. Do we need aprobation time to be allowed to continue working on U-Boot[ARM] again?
Nobody prevents you from working - just go on. The only restriction that _everybody_ will be facing is that CVS write access is not granted immediately. Send me your patches, and I promise to merge them in with the minimum possible delay. Then we'll discuss write permissions again. My descision is based on experience from the PPCBoot project, and I see no good reason to change my mind.
Best regards,
Wolfgang Denk
Hi,
On Mon, Nov 04, 2002 at 06:59:33PM +0100, Marius Groeger wrote: [...flamewar deleted...]
Ok folks, calm down everyone. I think that we have to find a solution which is acceptable for Wolfgang as well as for The Sysgos (TM). The situation is not easy, we all know that things havn't been developing optimally in the past, but having an unified tree now as a base for future work is definitely a beginning.
Looking from outside, I think the main arguments are these:
- Wolfgang definitely has the most knowledge about the project. Him having the best overview about the design, I can understand that he wants to stay in control about what goes in.
- Marius and Rob, being the ARMboot maintainers, naturally had the same "rights" on "their" project in the past and don't want to change this.
IMHO the only reasonable way is to _start_discussion_ on a u-boot developers' mailing list, so that larger design decisions are done only after having them discussed on the list. At the moment we only have a users' list. I'm thinking about something like the RTAI development model.
I understand that Wolfgang wants to avoind that code-that-doesn't- follow-The-Plan goes in, but you should take into account that if such things have happened in the past it was at a time where there _was_no_ team of developers behind the project who had to agree on a common solution.
Would it be a compromise when Wolfgang gives write access to Marius and Rob and they promise to do major changes only after having them discussed on the list?
Robert
On Mon, 4 Nov 2002, Robert Schwebel wrote:
Ok folks, calm down everyone. I think that we have to find a solution
Actually I'm surprisingly calm. Don't know why. :-) I must be getting old. I was hoping not to write flames yet. Sorry if it sounded otherwise.
[quite fitting analysis of status quo]
IMHO the only reasonable way is to _start_discussion_ on a u-boot developers' mailing list, so that larger design decisions are done only after having them discussed on the list. At the moment we only have a users' list. I'm thinking about something like the RTAI development model.
Hm.. Is that really a good reference? Honestly, RTAI tends to be a little bit chaotic from time to time. Also, I think multiple lists tend to be confusing, and even needlessly cut off users that might have valuable contributions to a discussion.
Would it be a compromise when Wolfgang gives write access to Marius and Rob and they promise to do major changes only after having them discussed on the list?
Yes, and that's all that we ask for. Well, not quite all: all maintainers have to give that promise, of course.
Best Regards, Marius
----------------------------------------------------------------------------- Marius Groeger SYSGO Real-Time Solutions AG mgroeger@sysgo.de Software Engineering Embedded and Real-Time Software www.sysgo.de Voice: +49-6136-9948-0 Am Pfaffenstein 14 www.osek.de FAX: +49-6136-9948-10 55270 Klein-Winternheim, Germany www.elinos.com
On Mon, Nov 04, 2002 at 07:44:50PM +0100, Marius Groeger wrote:
Hm.. Is that really a good reference? Honestly, RTAI tends to be a little bit chaotic from time to time.
It is - but take into account that RTAI is a _much_ larger project than U-Boot. But is there a better way? I especially like the way developers are dealing with each other in RTAI.
Let's face the truth: we will _never_ have a case where everybody is happy with what we have. There is _always_ compromise.
Also, I think multiple lists tend to be confusing, and even needlessly cut off users that might have valuable contributions to a discussion.
One list is ok for me, just let's start _discussing_ instead of crying after everything went wrong.
Robert
On Mon, 4 Nov 2002, Robert Schwebel wrote:
On Mon, Nov 04, 2002 at 07:44:50PM +0100, Marius Groeger wrote:
Hm.. Is that really a good reference? Honestly, RTAI tends to be a little bit chaotic from time to time.
It is - but take into account that RTAI is a _much_ larger project than U-Boot. But is there a better way? I especially like the way developers are dealing with each other in RTAI.
Couldn't agree more. Talking of reference models: how do you like the MTD project and list? Highly competent folks, fair access to the CVS system, well defined responsibilies.
Let's face the truth: we will _never_ have a case where everybody is happy with what we have. There is _always_ compromise.
Right.
One list is ok for me, just let's start _discussing_ instead of crying after everything went wrong.
Well - isn't it discussing what I'm trying to do?
Best Regards, Marius
----------------------------------------------------------------------------- Marius Groeger SYSGO Real-Time Solutions AG mgroeger@sysgo.de Software Engineering Embedded and Real-Time Software www.sysgo.de Voice: +49-6136-9948-0 Am Pfaffenstein 14 www.osek.de FAX: +49-6136-9948-10 55270 Klein-Winternheim, Germany www.elinos.com
Efficient Open Source development requires working along the "Release Early, Release Often" scheme. This is severely hampered if all updates are channelled through one person, who might not always be available.
In the KDE project one get's very fast a CVS account. The last time I counted them they were 767 people with CVS write access. It looks like that an open approach does not kill even big and complex projects.
participants (4)
-
Holger Schurig -
Marius Groeger -
Robert Schwebel -
Wolfgang Denk