Lets get inclusive(re: RFC, Coffee, etc)

[Rahul Dave <rahul@reno.cis.upenn.edu>]

Ok,

Lets try and get more constructive here.

Firstly, to those questioning Robert Current's dedication or intentions,
go see http://www.freshmeat.net/editorials/ and scroll down to the bottom,
and see who was one of the early evangelizers in the eyes of the Linux public..

Secondly, I do not think there is broad disagreement that some sort of
standardization of X libraries is needed. Some folks would suggest that
LSB is not the place for it.

Personally, I believe it needs to be layered. I have a smallish cluster of about56 nodes in a beowulf, 2/3rds of which have no X whatsoever, but each of which
can function when NFS fails as an independent node for a reasonable set of user
functionality. This involved removing cruft from usr, and recompiling a fair bit of RPM's into /usr/local, which is what I mount(not sure if thata the right naming, but its the right concept).

So, I dont need X. But, I need, say blast, 
which is a binary only genemoics software existing in multiple versions on
the cluster. Now they say rh5.2. What will happen, post LSB, is that they will
say LSB. And refuse to support me if I dont have the full LSB system on each
machine I run blast on. (I'm concocting the example but you get the point).

So LSB layer 1 ought to have only basics, and compliance ought to be in levels,
where compliance at a higher level means it at a lower level but not vice versa.
I'm sure that a few years down the line it will be useful to do some desktop
standardization, and this way we will build on previous layers of work.

I think layering is reasonable, and a inclusive solution to a non-easy problem.
Rahul