[Openicc] Helping with colord

[Alexandre Prokoudine]

On 3/8/11, Graeme Gill wrote:
> Alexandre Prokoudine wrote:
>> Cinepaint? Not updated in years, not communicating to users, not used
>> (comparatively).
>
> Hmm. How much money have you paid to use it? Enough
> to employ one or more a developer full time ? No ?
> Then perhaps it's a bit much to expect it to be developed
> and supported like a commercial product.

And thus we go back to good old "have you paid for it" as if it was
any sort of a valid argument.

OK, full reply goes below.

>> Having stamina for a long run is simply not enough. Having nice
>> architecture is not enough either. The only way to stop people
>> wondering if a project is dead or alive is to arrange matters so that
>> no sane person would ever have reasons to doubt. Of course if you
>> prefer getting architectural on people instead, it's entirely up to
>> you.
>
> It seems that you are deliberately conflating Cinepaint and
> Oyranos in terms of them being dead or alive. Again, hardly fair.

I'm giving you perspective of users. Just last week we had a mail from
a user to feedback at audacityteam asking if Audacity was dead, because
the last news on a new release was from April 2010. I personally was
kicked in the fork at osnews.com for not updating inkscape.org every
*week*. Was it fair? Hardly. But this is how things work. One can
accept that and work around it, or continue dreaming of a perfect
world to live in and not getting a life.

Do you know when Cinepaint's source code was officially released last?
Have a look yourself:

http://sourceforge.net/projects/cinepaint/files/CinePaint/CinePaint-0.22-1/

Do you expect it to work on modern systems flawlessly? Well, I don't,
because I tried that tarball. It kept crashing whenever I tried
opening Preferences dialog even two years ago, and I don't expect
newer systems to magically solve that. How fair is that towards users?
Should they be paing a developer to officially release at least a
patchset of everything that's been secretly happening in some Git tree
nobody know about? How much is for a new tarball? Maybe we should
start a pledgie to pay Robin/Kai-Uwe for running 'make build' and
uploading a file?

> If Oyranos's technical or architectural limitations are what is
> handicapping it, then by all means point out what you think they
> are.

This has already been discussed in the epic elektra thread. Do you see
any point getting back and starting it all over again?

> I certainly have my own reservations about it. But it's
> possible that some of it is purely opportunistic or structural,
> since someone being employed full time by Red Hat
> has a huge advantage in getting their work into wide distribution,
> irrespective of its underlying merit compared to the alternatives.

OK, here goes full reply.

Designing a perfect architecture and implementing it can work. The
prerequisite, however, is half a dozen of engineers on each side
working paid full time. I've been there, having had worked for an
enterprise company.

But when all you have is a handful of developers who work mostly in
their spare time, it's either perfect architecture that takes
virtually everything into consideration, or an actually ready to use
implementation of not so perfect architecture that will be fleshed out
later. Either, or. Period.

Do you want to see what happens when you go for a perfect architecture
instead of starting small and progressing step by step? Visit
lumiera.org. Excellent idea, shot in the head by overdesign. After
three years there are still no downloads, because there is nothing to
download. Not even an alpha is available. Do I want the only viable
desktop integrated color management app be shot in the head likewise?
Hell, no.

Is it possible to create anything complex with limited resources? Yes.
You start with a basic idea, provide rough yet working implementation
and progressively flesh out things, while adding other things. If you
want an example, darktable got from a newly started experimental app
to a working color managed digital photography workflow tool in just
two years with a handful of developers (just one the first year if its
existence, not counting myself). There is nothing special about not
shooting yourself in the foot, just keep the finger off the trigger
and don't point the barrel downwards (in fact don't point it anywhere,
keep it in the closet).

Back when we started OpenICC, the idea was to agree on concepts, work
out common solutions and then bloody well get things done. I love the
fact that we have best experts in CM here, as well as both proprietary
and free software developers and hardware vendor engineers. I love the
fact that I learnt tons of stuff. But it's been, how many -- four?
five? -- years since then, and the question I'm asking is: are we
getting things done?

Alexandre Prokoudine
http://libregraphicsworld.org