On Sun, May 24, 2015 at 3:48 AM, dmccunney <dennis.mccun...@gmail.com> wrote:
> On Sun, May 24, 2015 at 1:27 AM, Rugxulo <rugx...@gmail.com> wrote:
>> Dennis, once again, you're on the wrong mailing list if you're going
>> to constantly harp on how obsolete and useless DOS is.
> Please learn to read.  I never said DOS was useless - I said it was a
> legacy OS few folks have a real reason to care about.  Like it or not,
> that's true.

I think your perspective is far too limited to be declaring this as
absolute truth.

>>> People involved in things like compiler writing will be targeting Windows, 
>>> Linux,
>>> and the like.
>> Windows and Linux are too volatile to rely upon.
> Too volatile to rely upon?  If that's the case, perhaps you can
> explain why the vast majority of desktop systems run Windows,

The vast majority are "Home" users (not "Pro" or "Enterprise"), which
means they don't do "production" work (often considered only
multimedia hounds). And most of them, by far, apparently run (dead)
WinXP or (extended support only) Win7.

Heck, didn't you say you run Win2k?? That's far from compatible with
most modern apps. Most compilers won't even target it anymore. (Even
XP is better off, barely, for now. But at least you aren't trying to
run NT 4!)

I think you overestimate the amount of legacy (and compatibility) that
Microsoft is interested (and qualified) to support as far as end users
are concerned. It's not anywhere near 100%.

> and the vast majority of web servers run Linux?

All running the exact same kernel? Supported upstream? Drivers compatible? No.

> Many millions of people rely on both daily.  I'm one of them.

No, they don't rely on them, they use them. There's a big difference.
The world would not end if they had to reinstall or upgrade. If
anything, upgrading happens far too often (and is often mandatory,
breakage or not).

> Progress brings volatility with it, but do does life.  Somehow, we all manage 
> to deal with it.

Constantly destroying everything just to rebuild is not a sane way to live.

>> The fact that "nobody" (according to you, although I can name a few
>> non-commercial ones) targets DOS is irrelevant to us.
> Name whoever you like.  I'll be more impressed if you name any
> *commercial* offerings targeting DOS,

There are still many commercial compilers sold that target DOS.
However, I don't honestly know whether they are (strictly speaking)
"maintained" or not.

> since the reasons people care about things computing related  tend to involve 
> money.
> One of the reasons DOS gets little love these days is that there is little or 
> no
> money in it.

I've read a similar opinion from Walter Bright (Digital Mars) before.
I figure he ought to know.

Then again, what makes you think people would pay when they won't even
use what's free? We've got GCC, FPC, etc. If those aren't good enough,
then nothing is. (And no, it's not the technical fault of DOS for lack
of developers. If they can't be bothered, that's more of a cultural

> Who will *pay* to have DOS code written?

Are you seriously telling me you'd pay to have Clang ported to DOS? Go
ahead, email Christian Lattner, tell him how much you're willing to
hire him for. And make sure to tell him why OpenWatcom and DJGPP
aren't good enough.

>>> Even the embedded market is being taken over by things like 32
>>> ARM CPUs without the "real mode" issues involved in systems running 16
>>> bit Intel architectures and the bewildering variety of memory models
>>> DOS programmers had to deal with.
>> What are you referring to, the kernel? Sure, that uses 16-bit mode
>> because there's no major incentive to switch. Lots of other things are
>> using 32-bit flat model (since decades!), e.g. DJGPP and OpenWatcom.
> And if you don't have a 32 bit kernel, you jump through all sorts of
> hoops because of it.  Look at the fun involved in trying to run
> protected mode stuff under DOS.

What's the difference? It works. It's well documented. Dare I remind
you that DPMI was invented ... by Microsoft ... for Windows?? Intel
published and propagated the spec. It was standardized.

What makes you think that 32-bit systems programming is magically
easier than anything else? It's still dicey no matter which way you
look at it.

> There's *plenty* of incentive to switch, which is why most of the
> computing world *did* switch, and uses something *other*  than DOS.

There used to be tons of DJGPP apps, but once NTVDM started sucking
the big one (and not existing at all on AMD64), lots of developers
jumped ship. Without a common, reliable, and easy to use environment,
it was much harder for people to care. (I think it's obvious by now
that FreeDOS cares more than Microsoft.)

> But lets imagine for a moment that $DEITY works a miracle to order,
> and FreeDOS magically acquires a working 32 bit kernel.  What do we
> get?  We theoretically get an OS with a much larger address space, but
> it still single user and single tasking.  How much traction do you
> think it will get?

Larger address space than DPMI already gives us?? And see, even then
you still think it's useless. So why do you keep harping on 32-bit as
if it's some kind of panacea?

The point stands (not for technical reasons!) that people do whatever
the hell they want to do. Stop pretending that all decisions like this
are totally rational. They aren't! Just because someone doesn't like
DOS for emotional reasons doesn't mean it's not capable of performing
reasonable tasks.

>> Don't tell me that (optional) 16-bit support prevents anybody from
>> doing anything. We've had both for a long time, much longer than ARM
>> has been popular. Don't pretend that ARM is superior, it is definitely
>> not!
> I didn't say ARM was superior.  I simply said even the embedded world
> was shifting to 32 bit CPUs because it *could* - the hardware has
> become small, fast, and cheap enough.  There's little reason *not* to,
> and once you *have* the more powerful hardware, you start finding uses
> for it.  It's the same reason why I predict that all cell phones will
> increasingly be smart phone because they *can* be.  They will be
> powerful enough to run something like Android, and will.

Android? That has ridiculous version fragmentation? I like it and use
it (sometimes), but it's far from perfect.

Anyways, ARM is probably nice, I don't know. But just so you know,
Intel is not sitting on the sidelines. In fact, weren't you aware that
Microsoft's Surface Pro 3 (and Surface 3) both use Intel (x86)?? So
no, they don't "need" ARM at all!

> ARM gets extensive use in the embedded space because because it's more
> power efficient than X86 when the scarce resource these days is
> battery life,

Sure, my (ARMv7) Android tablet gets 6x better battery life (e.g.
Flash / YouTube) than my x86 laptop. It also does a lot less and has
less RAM and storage. And it's buggier and was obsoleted faster. But
it was also cheaper.

> and there is substantial OS and toolchain support,

Not at all. Almost everyone (for now) still treats x86 (i686 or AMD64)
as tier one with everything else far behind.

> but there are other players like MIPS also in the market.

No offense to them, but they aren't even a blip on the radar.

> (And note that as we get 64 bit ARM CPUs, we'll start seeing them in the 
> server room
> *because* of that power efficiency.  Think folks like Google and
> Facebook *won't* install ARM based rack servers by the thousands if
> they can drop their data center power bills by doing it?  If you do,
> think again...)

Servers are not the same thing we're talking about. Google can do
whatever it wants, but home users aren't ever going to run servers.

And Intel has insanely good fabs. Didn't they mass produce a 14nm cpu
recently? Maybe it was in the Surface, I can't remember.

> And increasingly, even embedded applications require 32 bit address
> space and multi-tasking.

I'm not sure what you mean by "embedded" here. Traditionally, while
there is some, I don't think they did require multitasking. True, I
think Intel did phase out their 80186 embedded cpus several years ago
though some clones may still be produced, dunno.

But, again, embedded is not something home users care about. It's an
entirely different field. You're trying to combine all of these
totally separate needs into one thing, and that will never work. They
don't target the same demographic, and they don't have the same
developers either.

> Tell you what.  Restrict yourself to *only* FreeDOS and things that
> run under it.  Do *all* of your computing that way.  Do all of your
> web surfing that way.  Tell me how you make out.

No, I need to boot up Win8.1 with gigs of RAM just to calculate 15^5
or edit a text file. (sarcasm)

> I know it *can* be done, because a few folks here seem to successfully
> do it.  I could not.  Too much of what I do requires capabilities that
> do not exist in DOS and DOS apps, and *can't*.

A poor carpenter blames his tools.

> DOS is fun to play with, but "play" is the operative word.  The *work*
> gets done elsewhere.

No work gets done at all! Too busy arguing! Why solve technical issues
when you can rain on someone's parade instead? Give me a break,
Dennis, conversations like this are a waste of time. Can't you
understand that it's (usually) not technical problems that prevent us?

"Well, everyone else goes with the flow, so I guess I'd better follow
suit. Windows 10 FTW!" (Have you upgraded yet? And if not, why not?
It's cheap enough. Everyone else is using it. Throw away all your old
machines, Dennis, don't be such a luddite. Resistance is futile. You
will be assimilated.)

One dashboard for servers and applications across Physical-Virtual-Cloud 
Widest out-of-the-box monitoring support with 50+ applications
Performance metrics, stats and reports that give you Actionable Insights
Deep dive visibility with transaction tracing using APM Insight.
Freedos-user mailing list

Reply via email to