Re: [PATCH] latest ash has broken 'echo' command

[Santiago Vila <sanvila@unex.es>]

On Mon, 25 Oct 1999, Raul Miller wrote:

> On Fri, 22 Oct 1999, Buddha Buck wrote:
> > > "Be liberal in what you accept, be conservative in what you generate".
> 
> On Mon, Oct 25, 1999 at 05:37:55PM +0200, Santiago Vila wrote:
> > This is ok in the general sense.
> >
> > However, for the unstable distribution there is another said:
> > 
> > "Be conservative in what you accept, so that you discover bugs and
> > policy violations in packages which are not so conservative in what
> > they generate".
> 
> This is flawed.
> 
> It's reasonable to do such things in experimental, and it's reasonable
> to treat unstable as a surrogate for experimental for a short time after
> the release of stable, but in the general case this is the wrong approach.

Well, I don't know what the "general case" is, but we have done this
several times.

For example, think about the dpkg --force-overwrite option turned off by
default in unstable (certainly, not just for a short time after the
release of stable).

Or think about the removal of the /usr/spool symlink. It was done during
the unstable stage. We would not have ever discovered packages using
/usr/spool/mail in a hardcoded way if base-files had been uploaded just to
"experimental".

I don't see that the current issue differs too much from these examples.

> This is very much the wrong aproach when we're nearing freeze.

Nobody knows for sure when we will freeze potato, so "nearing freeze"
sounds a little bit meaningless to me.

We are always nearing freeze, because time always go by.

> And, "fix the right problem" higher priority.

We did not know there *was* a problem until Herbert applied to ash the
rule of being conservative in what we accept. So the rule was useful
in this case because it helped to discover the problem.

Thanks.

-- 
 "2a11bfefc57fd9f05e647e298002be56" (a truly random sig)