[Gnu-arch-users] Re: arch and linux 2.7

[Pau Aliagas]

On Tue, 28 Oct 2003, Tom Lord wrote:

>     > From: address@hidden
>
>     > On Tue, Oct 28, 2003 at 09:18:27AM -0800, Tom Lord wrote:
>
>     > > However, I think we need to be cautious on a few points:
>
>     > > 1) We shouldn't presume to know more than we know.
>
>     > >   For example, it's one thing to say "we ought to improve some 
of the
>     > >   error messages" but another thing entirely to say "in order to
>     > >   promote kernel-project use of arch, we ought to improve some 
of the
>     > >   error messages."
>
>     > Wasn't this thread on; "if we propose arch for kernel; we should 
finish this>     > first or we don't stand a change..." ?
>
>     > That was wat I got from it anyhow..
>
> "Finish this first" is an underspecified concept.

I'm convinced that we are READY. But there are a few things that need to
be done in order to avoid confusion (rename commands now and document
everything in the tutorial and a good howto) and performance problems (no
one will find acceptable 20 seconds for a commit), 1 minute to copy a
pristine tree and several other glitches).

But my underlying message was that these issues are solvable right now.

> Given the unbounded resources, I'm extremely confident I could go on
> "finishing" arch for the next four decades.

I enjoy your sarcastic comments and the truth behind them :)

> A more practical goal than "finish" is "satisfy actual needs".

That was what I intended.

> Now, as reasonable as "clean up certain error messages" is as a goal,
> in general, "things zander thinks up to add to the list" is not
> "things that satisfy actual needs" and not at all obviously related to
> any essentially political goal wrt the kernel project.  There's
> overlap -- there's good reason to make the lists -- but let's not
> overinterpret them.
>
> One scenario that would make sense for the kernel might be:
>
>       a) Linus decides to move to a free software system.
>
>       b) Linus kvetches about error messages.
>
>       c) Since fixing the error messages isn't that big a deal,
>            it's certainly an option.
>
>       d) Someone decides to provide resources to make it happen
>            (whether unpaid-volunteer hacking or money).

We have three kernel hackers in the list using arch. And, allow me to say, 
quite representative.

> But this sentiment that "Well, we want the kernel, we just _can't_ go
> to the kernel with the error messages in their current state, hey
> let's work on error messages...."  --- it simply doesn't make any
> sense to me.

> [As a point of fact, I'm actually making progress on the groundwork to
> fix up the "contextless error messages" that you're talking about.
> The issue here isn't me saying "Nah, let's neglect those error
> messages" because I'm actively working on them.  The issue is the way
> you're relating that to the vague goal of "get kernel project design
> wins".]

Well, I personally consider more urgent the other issues in order to
spread arch's usage. There are piles of patches pending to be integrated
in the best SCM around. Kind of pardoxical. And it's not a critic.

Pau