[Simh] Simh on Windows - long startup delays and long shutdowndelays

[Eric Schmucker]

Kevin,

OS religion discussions are so completely tiresome and useless. I tried to
avoid it by carefully selecting my wording, but to no avail.

> Eric Schmucker wrote:
> > I just don't accept the argument that MS employees are sitting around
> > thinking of ways to lock programmers into their OS and tools. I don't
> want
> > to start an argument about this since it's very off-topic and I don't
> feel
> > that strongly about it. But I've seen many situations where MS made a
> good
> > technical decision that was best for Windows programmers and people
> > complained that it must have been for some evil purpose.
> >
> You should really read the comments made by the "wine"
> developers. For example, they have found several cases of
> documented function that call undocumented ones, just to
> include a delay of some kind. These are obviously there to
> make it harder to write competing software as fast as what
> Microsoft supplies. I wouldn't be surprised by anything
> those guys do.

Every time Microsoft does anything, people make comments about evil
motivations. It's easy to do and kind of fun, but not rational. I've heard
all the arguments and examples, and like I said before, I'm not going to
deny any of them. My point is simply that for every example you can state to
make your case against MS, I can site an example where they made the best
technical decision, but were criticized bitterly. Couldn't the "wine"
example you cite be explained by the fact that most programs contain bugs.
How easy it is to spout speculation when you can't back it up. You have a
belief system based on hearsay and speculation.

> > To a large degree our individual views on the broad range of
> technology out
> > there are influenced by the environment that we grew up on. I started
> on
> > RSTS/E (Basic+) and then VAX/VMS for many many many years. My whole
> view on
> > what makes sense architecturally was solidified from the examples in
> the VMS
> > system and library APIs.
> >
> And VMS is mostly dead now. Just try to find something
> useful about it, like the price of a TCP/IP license by using
> the phone. Make sure you have several days to
> waste while being transferred, put on hold, and being
> asked which version of Windows it's for.

I'm not sure why you mentioned that VMS is dead. It doesn't seem to relate
to my statement above it where I say each of us has a view of technology
that is largely grounded in the OS we grew up on. Unless you were just
slamming VMS by saying it's dead to discount the foundation that my opinions
are based on. If you don't know enough about VMS to know it strengths or
already have your mind made up, then why even talk about it. But there is no
doubt that the VMS system calls were architected by world class programmers.
C-RTL - not so much.

> > So my first exposure to the C RTL and then Unix certainly didn't
> impress me.
> > I would never use those kind of APIs if I had a choice. You can tell
> when
> > world class developers architected a system/API and when it was
> written by
> > beginners.
> >
> Yes, the beginners write bloated, slow interfaces, and
> the experts write fast, simple ones. Right?

An IO read API call that takes a pointer to a buffer that doesn't also have
a length limit is seriously flawed. If you get unexpected data on the
channel, you get corrupted memory and complete process failure. Anyone who
creates code like this is a seriously bad programmer. Even if it's for an
internal method where you control both sides, its still more likely to fail
in the future when other people make changes and can't spend the time
pouring through all its idiosyncrasies. This type of coding cannot be
justified.

Would you say that the JDK's IO classes are well designed? The features in
VMS's RMS SYS$OPEN call (available in the early 1980s) are similar to what
is in the JDK today. RMS provides a huge number of features, many database
like. The C-RTL IO calls aren't even in the same ballpark.

> > Ouch. I think I may be attracting anger from some people (Unix fans)
> and I'm
> > sorry for that. I just wanted to add my support to the idea that if
> you grew
> > up on an OS like VMS, avoiding calls like fopen() is second nature.
> >
> Yes, Unix programmers are always looking for people to
> hate. Whenever you see someone programming C, they
> are always filled with a burning hatred of everyone in the
> universe, living or dead. Their only thoughts are filled
> with the burning rage of how much they truly detest you.
> 
> Now, do we need any more of these constant "those Unix
> guys hate me, they really really hate me" paranoia messages?

Knowing that any discussion comparing OSes or programming environments can
trigger ridiculously silly responses, I tried to head that off by
apologizing and acknowledging that I might be doing it. Your response
clearly shows that my guess that this might happen was accurate. Not
paranoid at all. Your comment makes no argument or point, it is just a nasty
statement.

> > The fact that Simh runs on so many systems is quite a feat and I
> respect and
> > praise all the developers who make it possible. But I don't envy them
> trying
> > to use the C RTL calls whenever possible to make it portable.
> >
> You hate the Unix RTL, but are impressed by the large
> portable programs written using it. You like the VMS
> RTL, which hog-ties you to one specific line of hardware.
> Me, I've learned that portability is very important and
> useful, especially if you write something assuming that
> it isn't. It always comes back and bites me.

Our different views on this comes down to our background. You've worked in a
world where portability is important. I've never run into it and I've
programmed continuously for 30 years. The computer field is a wide one.
People who think their own experiences are universal are ignorant of the
wider world around them.

I acknowledged that the people working on Simh are working in an environment
where portability is important and I sympathized with the difficulty of
this. Others in this thread have made the point very well that Simh has a
lot of platform specific code in it. This is done when the C-RTL is limited
or buggy on one of the platforms. I said the C-RTL is primitive, not that is
has no place.

It seems to me like you were arguing the same old tired arguments instead of
actually reading what my points were.

> 
> _______________________________________________
> Simh mailing list
> Simh at trailing-edge.com
> http://mailman.trailing-edge.com/mailman/listinfo/simh