why learn UNIX

[Garry Wiegand]

In a recent article dalem at hpfcph.HP.COM ( Dale McCluskey) wrote:
>Two comments.  First, while UNIX manuals aren't designed with beginners in
>mind, they DO tell you a great deal that you will have trouble finding in
>VMS manuals - file formats, for instance.  

True, sometimes. For a counter-example, try Tar-file format - we just
replicated tar onto VMS, and it was surely a nuisance that that format wasn't
written down anywhere! 

DEC used to put more information in about file formats; nowadays it's more 
of a "need-to-know" - for example, the object-code language is documented 
heavily (you might be writing a compiler) but the disk home blocks are 
not ("why would you ever want to know *that*?")   Silly people - I want
to know *everything*.

By the way, the "VMS INTERNALS" book does have lots of good dirt in it. I 
think we bought ours in the bookstore down the street rather than from DEC...

>					   Second, UNIX is a fairly open 
>system that encourages experimenting.  This is aided by the information   
>available in the manuals.

I assume you're talking about the operating system kernel; there's been
times when would have *loved* to have a VMS kernel source license!
(hack hack hack...)

>It is also very  flexible.  An example of this is that one could write a
>shell that would run on UNIX and look like DCL (VMS's shell),  but you
>would have a pretty tough time doing the reverse.

Not true. I wrote a small Csh for VMS a couple years ago, but the exercise
got boring and I stopped. Csh is basically user-level C code, and at that
level the abilities of the two systems are nearly indistinguishable - the
major one is that VMS is slower to fork. That *is* a nuisance, but if you
are a little more clever, you can come out ahead on VMS - just cache and 
recycle the forked processes. (I've heard rumors of undocumented Unix 
"kernel hooks" to make the shell run fast; they might be something similar.)

garry wiegand   (garry%cadif-oak at cu-arpa.cs.cornell.edu)