Re: [gentoo-dev] RFC: Niches and new developers (especially xbox)

[Pieter Van den Abeele]

Hi,

The Xbox2 is rumored to be powered by 3 IBM 980 cpu's (dual core G5), 
same goes for the Playstation III. The current PS2 has a MIPS-SMP 
architecture. The current Xbox has on a slightly modified Intel 
architecture microprocessor.
All of them have very advanced GPU.

On 24 May 2004, at 19:48, Jon Portnoy wrote:

> I'm looking for comments on what, exactly, constitutes a platform worth
> recruiting developers for.

Historically we've used architecture/platform words in the following 
contexts:

- architecture: intel, ppc, power, mips, sparc, ... 32bit/64bit

- platform: windows, linux, bsd, cygwin, OS X, Solaris...

For the game consoles mentioned above, the question whether or not 
devrel can reject devs specifically for work on a game console, depends 
on whether the architecture in question supports the game console. 
Devrel should be able to impose extra requirements (in coordination 
with the mentor!!!) when the area of responsibility for the new dev 
includes only supporting the niche (or just one ebuild... things like 
that). The management is responsible for taking care of new 
architectures/platforms.

My opinion is that if there are people wanting to work on something 
that is legal to support, and that already may be supported by other 
distributions, why not? ( A new arch/platform needs to be discussed by 
the management. gentoo-alt provides help during the first 
days/weeks/months for new archs/platforms wanting the help ) Devrel 
should decide whether just supporting a niche market is enough to 
qualify as a new dev (needs to happen in coordination with the mentor).

I'm personally inclined not to support Xbox officially for legal 
reasons: It's not because you can't prove that something is not legal, 
that it is legal.

> I like to be extremely careful when selecting
> new developers both to keep the bar set high for developer status and
> for security reasons (every new account is a potential attack vector,
> especially when you consider that a developer's box could be
> compromised
> and used to get in), so I'm very inclined to reject requests for new
> developers

A QA team/tool monitoring contributions would help alot.

We already support a huge amount of niche markets. I don't believe the 
argument that 'devs for a niche market get rejected per definition' 
argument holds. I do agree that devrel can impose extra requirement for 
applications when the area of resp. for a new dev is too narrow (or too 
ambitious). All of this should happen in coordination with the mentor 
of course.

The security argument doesn't hold either (cfr. below).

> who will only be working on a very small niche like XBox
> support (especially when it's really just embedded x86 and it seems
> like there wouldn't be too much maintainance work involved). I would
> also be inclined to reject developers for, say, m68k.

Well, I have m68k machines in my basement. I doubt that a lot of m68k 
users would be interested in a from source approach. The only thing we 
need to support m68k users is a modified bootloader (floppy), and maybe 
some patched kernel ebuild. If somebody wishes to maintain that and do 
some ppc work as well, he's welcome.

I think that people with AMD64 machines allowing half the world to work 
on their fast machines pose a bigger security risk than one user 
ambitious enough to install gentoo on his favorite 33mhz 68K machine 
loaded with 24M ram and (maybe) a 512M hard disk.

Kindest regards,

Pieter Van den Abeele


--
gentoo-dev@g.o mailing list