Re: [Gnu-arch-users] Features command for arch

[Robert Anderson]

address@hidden wrote:

> On Thu, Sep 02, 2004 at 07:50:59PM -0500, Robert Anderson wrote:
>  
>
>
> [Features as a cached result of a config run]
>
> Actually, there's another thread proposing something similar
> (focussed from a slightly different --to me more appealing--
> point of view: features as assertions you check against actual
> tests at some test time).
Who writes these tests, and for what purpose?

It's a neat idea that configure tests could also be included in a test 
suite and vice-versa, but I think the idea that all possible things that 
you might want to ask about will have a pre-made test (by the 
developers?) is not realistic.  It's like the often talked about and 
intellectually pleasing idea that a test suite defines tla behavior with 
the implementation just being incidental.  Yeah, that's nice, but... IMO 
it's not going to happen.

> There's one thing I miss in this thread. Some day tla might be
> extensible. You might have to include a whole set of
> ``/usr/share/lib/tla2.5/pika'' trees in your md5-summing, right?
>  
That difficulty seems to apply equally well to any kind of cached result
for tla behavior.

I would imagine you'd want to look at the init files in that case, 
rather than
all the possible files the init files might load.

Bob