hey, remember the platform?

[Chris Lee]

Havoc Pennington wrote:
> On Fri, 2005-07-01 at 01:11 -0600, Aaron J. Seigo wrote:
> 
>>then let me rephrase for clarity: adding cairo right now is jumping the gun. 
>>it is too early for it to be a part of the platform release without 
>>redefining what the platform is about. let's revisit cairo as a candidate at 
>>a time when it makes sense, e.g. when there are stable pieces of software 
>>relying on it.
> 
> I agree with you on this point, but I will note that to my knowledge all
> of Mozilla, GNOME, and OO.org are more or less committed to the Cairo
> path. OO.org probably most weakly, GNOME probably most strongly. But I
> know there are sponsored developers driving Cairo for both GNOME and
> OO.org.

My other response to this thread was sent to jdub personally instead of
to the list by mistake, and I don't have a copy of it, but let me sum up
some of the points here.

GNOME has committed to Cairo, sure, that much is certain. And
OpenOffice.org may have a branch in CVS that uses it for chart and graph
rendering (so that we can sit through MORE useless bullshit charts and
graphs, but they'll be even prettier and smoother! woohoo.), but it's 
based on an API that is still evolving and is by definition going to 
break at least a bit before it's done.

Mozilla's devs are awesome, but they do have a history of dropping
support for changing APIs, especially when there aren't developers
committed to maintaining that support. As far as I'm aware, no Moz devs
are being paid full-time to maintain Cairo support, and the API is still
likely to change - possibly in major ways - before Cairo hits 1.0.

> My view is that we could have any piece of software that's used in
> several major projects in a fd.org platform, it isn't necessary for all
> projects to use it. Especially for something like Cairo, there's really
> no end-user impact if people choose to roll their own (unlike say
> fontconfig) so there should be no "pressure" to use Cairo as there is
> for something like fontconfig.

Not "all projects" should be using it, but if you don't have support
from both KDE and GNOME that's kind of a showstopper in my book.

And unfortunately, in spite of the good intentions of many of the
involved developers, I have no doubt that if we *do* let any software
not being used by both desktops into the platform, there WILL be
pressure on the desktop not using it to adopt it because otherwise
OMFGWTF THEY ARE REINVENTING TEH WEELZ.

We don't need that shit. We already have enough of it *without* a
platform release.

> Still as you say, we shouldn't be adopting pre-1.0 not-yet-shipping
> stuff.

Completely and totally agreed here.

As soon as Cairo hits 1.0 and there is an (even if it's just optional)
backend for Qt4's Arthur, I fully support including it in the platform.
Support from Mozilla and OO.o is just icing on the cake.

The thing is, if fd.o is to maintain any sort of relevance, we have to
be very careful about how we handle things like this. Our platform
shouldn't be controversial or exciting or even sexy. It should be boring
and obvious and 'duh' and it should make the lives of the third-party
developers easier. We want the platform to provide the answers to
questions like:

"How do I make my app show up in the menus everywhere?"
"How do I send a message to the rest of the system?"
"How do I do <foo> in a way that works everywhere?"

If people start trying to use things like the platform as leverage for
an argument and they piss off developers, fd.o loses. We don't need to
go back to the days when KDE and GNOME had completely different ways of
doing everything and nothing worked between desktops, but if we aren't
careful then people will just start ignoring fd.o and that is what will
happen.

-clee