[Info-vax] What would you miss if DECnet got the chop? Was: "bad select 38" (OpenSSL on VMS)

[johnwallace4 at yahoo.co.uk]

On Monday, 19 September 2016 08:16:45 UTC+1, Dirk Munk  wrote:
> Scott Dorsey wrote:
> > Dirk Munk  <munk at home.nl> wrote:
> >> Scott Dorsey wrote:
> >>> Dirk Munk  <munk at home.nl> wrote:
> >>>> Scott Dorsey wrote:
> >>>>> Jan-Erik Soderholm  <jan-erik.soderholm at telia.com> wrote:
> >>>>>>
> >>>>>> I think we have to accept that the rest of the world selected
> >>>>>> TCPIP for networking.
> >>>>>
> >>>>> And, that being the case, we need to have the same features that
> >>>>> people
> >>>>> have liked with DECNET (such as the remote save sets) available with > IP.
> >>>>
> >>>> YOU ALREADY HAVE THOSE FEATURES !!!!! IT IS CALLED DECNET OVER IP !!!!!
> >>>>
> >>>> Why on earth would any one try to invent something that is already
> >>>> there, that is plain silly. No other OS could use those features.
> >>>
> >>> There's the problem right there.  "No other OS could use those features."
> >>>
> >>> Maybe I want to put my remote saveset on a Solaris machine.  Maybe I
> >>> want to put it on a disk appliance.  We live in a world where we need
> >>> to coexist.
> >>
> >> You create a container file on any OS, and offer that container file
> >> over iSCSI to VMS. VMS will see a volume, you can format that volume
> >> with VMS, and use it to store data.
> >> The host operating system will only see a very big file, it can not look
> >> inside the container disk for individual VMS files..
> >
> > That doesn't sound very much like compatibility to me.
> >
> > Sheesh, just give me a standardized shared filesystem over IP, that I can
> > address from the command line.   With good performance.  I don't even care
> > what kind it is as long as it's standard and widely-compatible.
> 
> How do you want to achieve that? VMS has file versions, Unix and Windows 
> don't. Do you want to abandon VMS file versions, or should Unix, Linux 
> and Windows learn hows to use file versions?
> 
> For Unix and Windows a file starts at point A and ends at point B, and 
> in between are bytes. If these files have record separators, then it 
> will be a <cr> for the one, and a <lf> for the other. VMS knows these 
> files as stream files, or files with an undefined contents. Normal VMS 
> files are structured, a concept completely unknown to Unix and Windows. 
> Cobol for instance needs structured files, so a Cobol compiler on Unix 
> or Windows has to define its own structured files, but Unix and Windows 
> will be completely unaware of that.
> 
> How do you want to combine these things in one file system?
> 
> >
> > There was a time when DECNET was a great thing.  Back in those days, my
> > employers ran a network with over a thousand DECNET notes in several
> > areas.  We were on NREN so that we could readily copy a file over to a
> > university in Spain or vice-versa.  It was great.  If anything went wrong,
> > I could call the networking office and they would fix it because everybody
> > there knew DECNET.
> >
> > The last two DECNET nodes shut down a couple years ago, after a decade or
> > so of being tunnelled through IP and a decade of my having to listen to
> > the networking guys about "those weird-ass servers."
> >
> > It was here, and it was good, and it was widely compatible.  But it's not
> > here any more, it's not good any more, and it sure isn't compatible with
> > anything much any more.  There was a time when it would have been possible
> > to extend it and keep it living, but that window closed long ago.  It's
> > time to leave it dead and stop wasting time, energy, and money that could
> > be spent taking care of the living.
> >
> > It's time to let it go along with SNA and Appletalk...
> > --scott
> >

Interesting isn't it.

Shiny sells, but engineering details make things actually
work (or, quite frequently in IT, not work) right.

Engineering by PowerPoint, as the NASA Columbia inquiry
(and various others since) called it. Others call it
faith-based engineering.