Re: Why doesn't everyone use Modules ??

[<re...@me...>]

> From: "J. Pelan" <J....@Qu...>
> 
> I was recently made aware of the Modules package when Cray 
> adopted it for use with UNICOS and their mutifarious versions
> of compilers etc.

Cool.  I'm glad to see a vendor using it :-)

> I appears to be an extremely useful tool and one which I'm sure
> many people have thought about developing from time to time.
> Seeing as it is so good why aren't more people aware of it ?

Well, the main reason is that it fixes a problem which is mostly
peculiar to large sites: how to handle user environments with
lots of applications.  Most folks are stuck in either the PC
space (few applications, no concept of multiple users) or the
mainframe space (few applications, many users.)  They have yet
to tackle environments where you have heterogenous systems, 
many applications, many users and applications which change 
over time by adding new versions.  IMHO the latter is the most
important benefit of a Modules environment.

> It doesn't seem to have made an impact in any FAQ (like
> comp.unix.shells for example) and indeed it took me a while
> to find the Tcl implementation using AltaVista.

The primary vehicle has been USENIX and LISA conferences.  LISA 
is particularly important because that is the Large Installation
System Administration conference.

> So where's the catch ?? 

The catch is that a single user system, like most Linuxes, don't
really need Modules: it is extra overhead.  Most large sites are
either flailing, using Modules, using something else like Depot 
or rolled their own.

The big question is: "what does the future hold?"  I've been 
mentally struggling with this for some time.  There was a time 
when a univeral GUI like CDE would be the answer.  With CDE there
is enough administrative flexibility to solve many of the problems
Modules solves for GUI environments.  But the new universal GUI is
really a browser.  And the web, especially with Java applets, has a
new way to solve many of the user environment and application 
distribution problems.  Alas, we still have the problem of non-GUI
environments, so there still isn't a clear winner.

 -- richard