[clug] Systems Software Research Irrelevant

Rob McAuliffe rmcauliffe at ozemail.com.au
Thu Jun 17 21:39:43 GMT 2004


>http://cm.bell-labs.com/who/rob/utah2000.pdf
>
>Have many of you seen this piece?
>Do you think any/all/some of it is still valid??
>
>Is it just sour grapes from someone who's pet project (Plan 9) didn't do
>as well as they wanted.
>  
>
...

>Steve Jenkin, Unix Sys Admin
>  
>
I can't say I'd seen it before.  It would seem to me that the comments 
are somewhat valid, however there seems to be a bias towards "OS and 
languages are interesting systems" ignoring other parts of IT.

So maybe its valid to some extent, but as a counter-example: at the ANU 
there is a virtual reality theatre that is used for research projects 
(called the Wedge).  Just as OSs were fledgling in the days he refers 
to, and small projects could contribute interesting ideas, so also are 
interfaces and systems in VR.  So a small project *can* contribute 
something innovative and interesting.

OK, so its a little 'down the track' (although there *is* a trend for 
current destops to incorporate more 3D graphics), but I think it 
classifies as 'interesting'. 

Or maybe I'm just biased because I'm doing an Honours thesis on it :0)

So maybe the interesting research has moved up the stack, leaving this 
guy behind.  Maybe the "how do I efficiently interface with hardware" 
problem that OSs solve has been solved to enough peoples satisfaction to 
require no further evolutionary improvements, meaning that the 
interesting evolutionary improvements are 1 or more steps up the stack: 
new systems and ways of working sit atop the previous innovative platforms.

Web services is one example of this kind of thing.  A remote protocol 
that sits on top of HTTP.  Don't care about transmission, that problem 
has been solved, so just sit on top of it.

Mind you, that plays to some extent into his argument because (to my 
knowledge) most of web services was spawned by MS :)

Rob McAuliffe


More information about the linux mailing list