[Development] qtchooser (was: Re: Adding new third party component three.js to Qt?)

[Thiago Macieira]

On Sunday 18 January 2015 06:08:47 Kevin Kofler wrote:
> Thiago Macieira wrote:
> > Unanimity was not required. Only consensus and that was achieved. The
> > requests for renaming were argued and heard (I argued for them myself, I
> > even prepared patches and had the solution in progress) but in the end the
> > consensus was another way. Now everyone, including me, has to abide by
> > that decision, unless there's new relevant information that didn't exist
> > then. You haven't presented any new information.
> 
> Of course I have. I pointed out several fatal flaws inherent to qtchooser's
> design, for which the only fix is to drop qtchooser entirely.

Please list them again, so we can address them. The one I can remember now is 
the default config file for multilib, which is fixable by using alternative. 
Multilib development is covering the sun with a sieve, so pardon me if I don't 
give it too high a priority.

> > Once again: I argued against it in 2011-2012, but you don't see me
> > rebelling against that now.
> 
> It was a mistake that you didn't fight harder.

I did as hard as was allowed. Our governance model is not "last man speaking 
wins". I had to concede when I found myself won out.

> I also see how it happened: You were convinced that it was possible to solve
> the problem in a different way (qtchooser) and so accepted to implement
> that. Unfortunately, that implementation does not fulfill the
> distributions' actual requirements, so we are back to square one.

I beg to differ. It solved the problems of the distributions that bothered to 
help me with the development.

> Now I'll also take some blame, because I was aware of several issues with
> qtchooser right back when it was introduced (but not all, e.g., I didn't
> know about the multilib thing), I voiced my dissent in several places (at
> least #fedora-kde IRC and the Fedora kde mailing list) and was trying to get
> others (mainly Rex Dieter and Sune Vuorela) to forward my objections
> (basically the ones I wrote down in this thread) to you, but this never
> happened (they wanted me to sign up to this mailing list and post them
> myself), and I didn't want to go through the usual lengthy subscription
> process (subscribe in mailman, act on the confirmation mail, disable mail
> delivery, subscribe the Gmane gateway to Knode, send the message, reply to
> Gmane's confirmation mail) just for that.

Since they were never sent, we apply the principle if "silence is consent". 
Therefore, you implicitly agreed with the implementation.

Now we have a legacy to keep, so we can't accept a radical change. Only 
incremental improvements.

-- 
Thiago Macieira - thiago.macieira (AT) intel.com
  Software Architect - Intel Open Source Technology Center