[Beowulf] small-footprint MS WIn "MinWin"

Jim Lux James.P.Lux at jpl.nasa.gov
Mon Oct 22 16:29:21 PDT 2007

At 02:01 PM 10/22/2007, Peter St. John wrote:
>I think we agree on some things (e.g., modular design is not 
>conducive to Digital Rights Management) and I must defer to your 
>expertise on most of what's left, but I have to speak up on a couple.
>On 10/22/07, Jim Lux 
><<mailto:James.P.Lux at jpl.nasa.gov>James.P.Lux at jpl.nasa.gov> wrote:
>I don't know that Windows (at least since NT) isn't actually already
>a small kernel surrounded by (dare I say "embraced and extended by")
>a lot of utilities.  Sure, they're not command line utilities with
>cryptic 2 letter names
>It's trivial to overlay aliases on any cryptic two-letter commands 
>you like. Fortune offered a menu-driven user interface for folks who 
>didn't like "creat" or "su" in Sys V, in '81 ish.

Sure.. but the basic design model of kernel surrounding by cluster of 
little utilities is how NT started out.  It has, as several have 
pointed out, morphed, for performance reasons, among others.

>and a man page full of switches.  However, an
>awful lot of what people talk about as "Windows" isn't the kernel (a
>lot of the GDI, for instance, has been separate from the "kernel",
>per se, since pre WinNT)
>I misread that, at first, as "GUI". I see that MS has the term 
>"Graphics Device Interface" and I'll allow that is separate from the "kernel".

Was separate, now part, now not.  It sort of changes, as graphics 
cards change and the "preferred coding model" changes (DirectX, etc.).

The game industry is responsible for a lot of the intertwining in 
Windows.  With the nice abstracted interfaces, they couldn't get 
enough performance, so they did things like pull some graphics ops 
into the kernel mode, provide all sorts of "hooks" and "bypasses" to 
use the accelerators, etc.

Linux doesn't have a huge gamer user base, and there's no single 
point that a game publisher and video card mfr could go hammer on to 
say "give me speed!", so there's no tendency to try and shortcircuit 
the architecture.

>It does all this now. The question is whether you can get rid of a
>lot of the other stuff, since at a very fundamental level, windows
>follows an event driven model, where the events are largely from user
>Ding. The "event driven model" that is "fundamental" in MSWin (since 
>W95 or W98) comes from the GUI (but not the GDI) being integrated 
>with the OS. Which is not conducive to interoperability.

But.. in the original NT (3.x) it was still separated, and you could 
actually run NT pretty much headless (I did this for some remote site 
applications).  I think over the years, it's gotten worse, because of 
the demands to dispatch the events more quickly.

>For a long time, the only way to remotely administer a NT box was 
>with a third party application pushing the pixels of the remote GUI 
>over the phone line to the troubleshooter's client. To me, using MSW 
>as a server was ...not efficacious. Programs don't talk to each 
>other using bitmaps and mouse clicks, unless there is a layer of OCR 
>in between. Imagine replacing every occurance of "|" is a 
>shellscript with a call to OCR AI.

I used to remotely administer NT 4.0 boxes with essentially only 
command line utilities.  The BORK was my friend.. it provided lots of 
command line equivalents to the GUI management utilities.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.beowulf.org/pipermail/beowulf/attachments/20071022/ca72fd0e/attachment.html>

More information about the Beowulf mailing list