On 2/03/2009, at 10:30 PM, Martin Bochnig wrote:
>> You can't have a set of technical decisions made by a referendum.  
>> Lots of
>> people without a strong vested interest in the code would  
>> essentially be
>> arbitrarily voting on what the 'perceived' best solution is,  
>> without any
>> strong personal research on it.
>
>
> Sorry Glynn, but this is always the case.
> Especially in real national politics.
> Folks will always elect what or whom their favorite TV station
> instructs them to (and most folks lack any halfway sophisticated
> political background).
> Do you then see this as sufficient to abandon democracy?

Yes, but design by committee doesn't work

http://en.wikipedia.org/wiki/Design_by_committee

>> There's very few open source communities out
>> there that would do such a thing.
>
>
> I am also on a few other devel lists (e.g. qemu, xorg).
> While there aren't actual referenda there, those discussions are
> indeed happening (openly / unbiased-ly).

Discussions may be happening, but ultimately very few involved in the  
discussion will be the ones who make the decision.

>> Quite often, developers gather around
>> something that they find most interesting, and momentum gathers  
>> from there.
>> It just so happened that more developers found disadvantages of  
>> conary
>> versus writing IPS from scratch.
>
>
> Really? Where can I find that specific discussion somewhere in the  
> archives?
> AT LEAST NOT IN PUBLIC. Maybe you have more knowledge than I, simply
> because you know how it looks on the other side of the firewall.
> Somebody inside SMI had made this decision "for" the community. Namely
> in advance (circa summer 2007). Period and end.
>
> Also, I heard it so often, that conary was so different and had so
> many disadvantages (and would therefore simply not be the right choice
> for OpenSolaris). This argument - as is - always gets fired back to
> me. Could you be a little more specific please? What exactly makes
> conary a bad choice?

http://mail.opensolaris.org/pipermail/pkg-discuss/2007-September/000048.html

Stephen touched on this back at the inception of the IPS project.

> And if it is really so different from what OpenSolaris needs, why is
> IPS cloning its functionality? Why is it even implemented in the
> similar framework (python2.4 cli front-end, shared C libs backend,
> sqlite3 database [and stubs for other db's] repo management)?
>
> IPS is a conary-clone.
> Without the ability to build packages from src.
> Don't you believe me, if so, why _exactly_?
>
> So, Sun's default argument that conary would not be what OpenSolaris
> needs, cannot be valid, because in that case why would you be cloning
> it from scratch, in a huge effort?

I'm sorry, but I can't speak for the IPS developers here. Only they  
can know the design choices they made.

>> If developers with runs on the board in
>> terms of contribution, trust and experience had gathered around  
>> conary, then
>> things might have been different.
>
>
> It was due to my - then - very poor style of interacting with the
> community and Sun.
> But what counts more, the best technical intentions and resulting
> benefits for our overall project, or some social problems due to
> untalented behavior, misunderstandings and misinterpretations?

I'm just suggesting that if you had been a core ON contributor for  
several years, things might have been different, that's all.

>

Glynn

Reply via email to