Re: RFC

[Stuart Anderson <anderson@metrolink.com>]

On Wed, 15 Mar 2000, Robert W. Current Jr. Ph.D. wrote:

> I am saying, there is a base.  Add X as a "layer."  Stuart did well in
> discribing it as "middleware," and I partially agree with him.  But I
> think in his discription, he isn't seeing that "base" and "middle" are
> layers, and IMHO, the LSB should be the "base" that the "middleware" sits
> on.

Maybe I didn't describe them adequately, but I very much seem things as
layers. I do however, think that it has been the intent of the LSB to
include common layers. The specification document has evolved some to try
and seperate these thing out into "layers" I will think about it some more
and see if it can be done (and makes sense to do so ) even more clearly.

There has been some recent discussion about taking subsets of the LSB for
embedded systems. What you are calling the base sounds like it may correspond
to one of these subsets.

We had been considering "layer 0" to be inclusive of lots of stuff. The
embedded subsets would be lower that "layer 0". It would be possible to
change "layer 0" to correspond to a more minimal base, and what we have
been working on would be a "layer 1" or "layer 2".

This wouldn't invalidate any of the work we have done so far, but it does
conflict with the goal of having "LSB compliant" be a single term that
means everything that a large (probably GUI based) application needs will
be present. It would be more confusing to call something
"LSB layer N compliant".

This whole discussion does however remind us to make sure we are working
from the inside out (or botton to top) when developing the test suites and
sample implementations.
 
> As well, I would say, what of Commercial X?  The base is LSB-Linux because
> all software is common, but if you were to pull out one X variant and put
> in another (MetroX? XiGraphics?) then how can you say it's the same base
> set when it clearly is not.

The LSB is a behaviour based specification. In the case of X, there is a
well defined behaviour, and a very thorough test suite already available.
I consider X to be "more stable" than the rest of what we are trying to
specify, which often lacks a written specification, or an adequate test
suite.

In the Linux market, the commercial X vendors provide X servers, not the
base X libraries. The base X libraries come from XFree86, and no one changes
those out.

                                Stuart

Stuart R. Anderson                               anderson@metrolink.com

Metro Link Incorporated                          South Carolina Office
4711 North Powerline Road                        129 Secret Cove Drive
Fort Lauderdale, Florida 33309                   Lexington, SC 29072
voice: 954.938.0283                              voice: 803.951.3630
fax:   954.938.1982                              SkyTel: 800.405.3401
http://www.metrolink.com/