Re: [PDO] PDO2 as a set of PECL extensions, rather than PHP core?

On 28/01/2008, Lukas Kahwe Smith <[email protected]> wrote:
> On 28.01.2008, at 22:20, Dan Scott wrote:
> > On 28/01/2008, Lukas Kahwe Smith <[email protected]> wrote:

<snip>

> >> Copying them into core before a release also seems absurd. If php.net
> >> ever determines to make some changes, there is no way to contribute
> >> them back (without a CLA) and as a result things will lead to a fork,
> >> especially if there are ever differences about implementation and
> >> code
> >> management. This is the show stopper .. not the current process of
> >> symlinking things from PECL into core.
> >
> > There are many absurdities, some more absurd than others. So let's
> > explore this one. You are positing a hypothetical event ("if php.net
> > ever determines to make some changes") and using an absolute ("there
> > is no way to contribute them back"). I don't believe that the absolute
> > applies to this hypothetical event.
> >
> > If, in fact, the a non-CLA change was made to the core code, a clean
> > room process could be used to communicate that change back to the CLA
> > project. This process doesn't have to be onerous; it could something
> > as simple as "we found that the PDOConnection objects were leaking
> > resources if they were closed without an explicit commit() or
> > rollback()".
> >
> > This could be described by someone who had not signed the PDO CLA,
> > assuming that there was someone in the PHP community who was willing
> > to try to help the overall PDO v2 project succeed and willing to meet
> > the participants on some middle ground, rather than taking an extreme
> > position one way or another.
>
> Did you mean to only reply to me? If not then feel free to reply to me
> with a CC to the list. I left your full quote intact so that the full
> conversation would be available in this case.
>
> The likelihood of changes having to be applied are fairly big. Now
> usually these changes will not be very significant, so that someone
> that has signed the CLA and is not afraid of being tainted or whatever
> could communicate them fairly easily. However I do not think that its
> unlikely that there could be fundamental differences in opinion about
> things at some point. Now this could result in a fork and things move
> on along separate paths from then on, no harm done to have some code
> developed until then.
>
> This of course assumes that each vendor essentially implements the
> entire driver on its own with no community interaction besides from
> those that do sign the CLA or where some proxy takes the time to
> explain the change so that it can be implemented in a way that does
> not violate any copyright. The patent protection offered in this
> situation would still be laughable. Also I want to make it clear once
> more, the second php.net decides its necessary to do greater changes
> the likelihood of a fork is considerable (unless that proxy finds a
> way to explain all those changes.  And of course the developers of
> vendor X agree to implement all those changes in a way that is
> acceptable to php.net without too many round trips to sufficiently
> piss of php.net or vendor X developers.
>
> Now all of this CLA mumbo jumbo obviously does not prevent
> discussions. However the fact that the specs are even to be CLA'ed off
> does make me a bit cautious to really have much of a chance to have an
> open discussion with any of the developers from those who require a
> CLA. And all of this is assuming that all the vendors can be convinced
> that the specs do not need to be handled by a CLA.

(Sorry! I realized after walking home that I never checked for your
reply on #postgresql, so I found it in scrollback).

You're right, I originally meant to reply to the list.

My goal was simply to try to find some middle ground so that PHP can
make further progress towards a good + standard database interface.

Your response, posing the worst possible outcomes and the use of
phrases like "laughable" and "CLA mumbo jumbo" suggests that my trying
to find some common ground is not a worthwhile use of my time. Which
is fine. I have lots on my plate.

What I see is one side saying: "Our lawyers are holding us back from
participating more openly because they want all participants to sign a
CLA that makes explicit what is the understood, but unwritten,
agreement amongst contributors today."

I see the other side saying: "We are insulted that you are asking us
to sign something that makes explicit what is the understood, but
unwritten, agreement amongst contributors today."

I don't see any "winning" outcomes in this scenario. Maybe the problem
is that the participants, particularly those on the PHP community
side, see this as a battle to win or lose rather than a discussion
that could lead towards some common ground. The biggest loser in this
stands to be PHP itself.

It's disappointing, and all the effort spent fighting this CLA
proposal could have been far better spent discussing the details of
what PDO v2 should become.

-- 
Dan Scott
Laurentian University