Vinnie Falco wrore:

On Tue, Aug 6, 2024 at 5:24 AM Peter Dimov via Boost mailto:boost@lists.boost.org > wrote:

I don't think this is necessary. Historically, outsourcing our decision making to some external entity has never worked, but even if it did, I don't see why we'd need a registered nonprofit for that.

Someone has to do it since the Foundation will no longer be responsible.

Why? The Foundation has never even tried to do this, and the previous such attempt (the infamous CMake announcement) has been a tremendous success.

On Tue, Aug 6, 2024 at 8:34 AM Ion Gaztañaga via Boost wrote:

El 06/08/2024 a las 15:13, Peter Dimov via Boost escribió:

...

Vinnie Falco wrore:

...

On Tue, Aug 6, 2024 at 5:24 AM Peter Dimov via Boost mailto:boost@lists.boost.org > wrote:

I don't think this is necessary. Historically, outsourcing our decision making to some external entity has never worked, but even if it did, I don't see why we'd need a registered nonprofit for that.

Someone has to do it since the Foundation will no longer be responsible.

Why? The Foundation has never even tried to do this, and the previous such attempt (the infamous CMake announcement) has been a tremendous success.

If the directional decision needs some shared resources, then the "entity" somehow needs to participate.

Also, this type of "directional decision when consensus cannot be reached" feature comes from the Steering Committee times (which was formed in 2011), so we can't say it's not in the Boost tradition:

https://web.archive.org/web/20141224160315/https://sites.google.com/a/boost....

However I agree that the outcome of such "directional decisions" is mixed. It failed with CMake, but we can say that SVN to Git migration (https://lists.boost.org/Archives/boost/2012/05/193493.php) decision was better handled.

The "directional decision" aspect IMO was never successful when it came to non-infrastructure decisions. I know this contradicts your examples, so let me explain the examples: The svn->git transition worked like this: * A couple of library developers implemented the process for doing the transition and provided a working model. * There were discussions on the list about that and the idea of the transition. * The community agreed to go to the committee and ask for a "thumbs up" to fully complete the transition. The cmake incident went like this: * There were lots of recurring discussions about cmake (the first one starting during the svn->git transition). * A couple of library developers started experimenting and implementing support of cmake for end users. * A couple of different developers, unaware (AFAIK) of that preliminary work, asked the Foundation to decide that Boost should move to cmake. AFAIK they didn't define what moving to cmake meant. But not sure as all that happened in the isolation of C++Now. Can you spot the differences? The successful case was to reaffirm what the community had already decided. The "unsuccessful" case was to usurp the community process. -- -- René Ferdinand Rivera Morell -- Don't Assume Anything -- No Supone Nada -- Robot Dreams - http://robot-dreams.net

On 06/08/2024 14:51, René Ferdinand Rivera Morell via Boost wrote:

The svn->git transition worked like this:

* A couple of library developers implemented the process for doing the transition and provided a working model. * There were discussions on the list about that and the idea of the transition. * The community agreed to go to the committee and ask for a "thumbs up" to fully complete the transition.

I wouldn't remember it that way. My memory is that Dave asked for the help of a select few in writing the tooling to do the conversion because the Boost community had proved singularly useless in taking a decision, and he was going to impose a decision as BDFL after a great deal of wailing and complaint about there being any change at all. He then rammed it through fairly unceremoniously giving people very little choice about it.

The cmake incident went like this:

* There were lots of recurring discussions about cmake (the first one starting during the svn->git transition). * A couple of library developers started experimenting and implementing support of cmake for end users. * A couple of different developers, unaware (AFAIK) of that preliminary work, asked the Foundation to decide that Boost should move to cmake. AFAIK they didn't define what moving to cmake meant. But not sure as all that happened in the isolation of C++Now.

Can you spot the differences?

The successful case was to reaffirm what the community had already decided. The "unsuccessful" case was to usurp the community process.

Another way of looking at that is today Boost has almost complete cmake build support. In terms of **eventual outcome**, it has been quite the success, and the Dimov got that done incrementally and non-coercively. I'd therefore say the second case example is the more productive and less dysfunctional personally. Niall

On Wed, Aug 7, 2024 at 4:51 AM Boris Kolpackov via Boost wrote:

Niall Douglas via Boost writes:

...

I wouldn't remember it that way.

My memory is that Dave asked for the help of a select few in writing the tooling to do the conversion because the Boost community had proved singularly useless in taking a decision, and he was going to impose a decision as BDFL after a great deal of wailing and complaint about there being any change at all. He then rammed it through fairly unceremoniously giving people very little choice about it.

Hm, that's not how I remember it, and I was on the Steering Committee at the time.

There was a heated Future of Boost session at C++Now where Dave used some strong words to express his frustration with a few Boost developers holding hostage the modernization and thus the future of Boost (or something to this effect). However, it was the Committee that then voted in favor of switching to CMake (and, no, Dave was not on the Committee).

(FWIW, I abstained from that vote due to what I thought could be viewed as a conflict of interest: I would have voted against and at that time I was already working on build2, so it could have been viewed as me trying to undermine the competition. But, looking back, I should have voted anyway.)

Niall was referring to the svn->git conversion. And, I think, you are referring to the cmake work? -- -- René Ferdinand Rivera Morell -- Don't Assume Anything -- No Supone Nada -- Robot Dreams - http://robot-dreams.net

On 07/08/2024 13:41, René Ferdinand Rivera Morell via Boost wrote:

On Wed, Aug 7, 2024 at 4:51 AM Boris Kolpackov via Boost wrote:

...

Niall Douglas via Boost writes:

...

I wouldn't remember it that way.

My memory is that Dave asked for the help of a select few in writing the tooling to do the conversion because the Boost community had proved singularly useless in taking a decision, and he was going to impose a decision as BDFL after a great deal of wailing and complaint about there being any change at all. He then rammed it through fairly unceremoniously giving people very little choice about it.

Hm, that's not how I remember it, and I was on the Steering Committee at the time.

There was a heated Future of Boost session at C++Now where Dave used some strong words to express his frustration with a few Boost developers holding hostage the modernization and thus the future of Boost (or something to this effect). However, it was the Committee that then voted in favor of switching to CMake (and, no, Dave was not on the Committee).

(FWIW, I abstained from that vote due to what I thought could be viewed as a conflict of interest: I would have voted against and at that time I was already working on build2, so it could have been viewed as me trying to undermine the competition. But, looking back, I should have voted anyway.)

Niall was referring to the svn->git conversion. And, I think, you are referring to the cmake work?

I was referring to the svn->git conversion. Apologies, I should have clarified I meant Dave Abrahams not David Sankel. Niall

On Sat, Aug 17, 2024 at 11:50 AM Dave Abrahams via Boost < boost@lists.boost.org> wrote:

...It should all be in the public record, regardless...

This brings up an interesting topic. In the last several weeks I've been studying the public records and it is not always clear to me the order of events or the exact dates. For example, was Boost founded in 1998 or 1999? I only started contributing in 2017 so I have no memories to call upon here. A personal project of mine is to construct an unbiased and accurate historical record of Boost for people who have joined the project more recently to understand the context of things and also the rationale for why things were done a certain way (or in some cases to understand why certain decisions may not have worked out so well later). I guess what I'm saying is that it would be nice if some of the O.G. Boost folks could sit for an informal interview and help offer more detail on the early years of the project. Thanks