Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > comp.os.linux.misc > #95569 > unrolled thread

Why macOS Will Never Be A High-Performance OS

Started byLawrence D’Oliveiro <ldo@nz.invalid>
First post2026-11-12 20:13 +0000
Last post2026-11-21 01:05 +0000
Articles 20 on this page of 22 — 13 participants

Back to article view | Back to comp.os.linux.misc


Contents

  Why macOS Will Never Be A High-Performance OS Lawrence D’Oliveiro <ldo@nz.invalid> - 2026-11-12 20:13 +0000
    Re: Why macOS Will Never Be A High-Performance OS Alan <nuh-uh@nope.com> - 2026-11-12 14:32 -0700
      Re: Why macOS Will Never Be A High-Performance OS The Natural Philosopher <tnp@invalid.invalid> - 2026-11-13 10:35 +0000
    Re: Why macOS Will Never Be A High-Performance OS CrudeSausage <crude@sausa.ge> - 2026-11-12 17:15 -0500
    Re: Why macOS Will Never Be A High-Performance OS Richard Kettlewell <invalid@invalid.invalid> - 2026-11-13 19:24 +0000
      Re: Why macOS Will Never Be A High-Performance OS Rich <rich@example.invalid> - 2026-11-14 04:02 +0000
        Re: Why macOS Will Never Be A High-Performance OS The Natural Philosopher <tnp@invalid.invalid> - 2026-11-14 12:07 +0000
      Re: Why macOS Will Never Be A High-Performance OS The Natural Philosopher <tnp@invalid.invalid> - 2026-11-14 12:06 +0000
      Re: Why macOS Will Never Be A High-Performance OS WolfFan <akwolffan@zoho.com> - 2026-11-16 15:43 -0500
        Re: Why macOS Will Never Be A High-Performance OS Nick Charles <none@none.none> - 2026-11-16 21:21 +0000
    Re: Why macOS Will Never Be A High-Performance OS Lawrence D’Oliveiro <ldo@nz.invalid> - 2026-11-18 04:43 +0000
    Re: Why macOS Will Never Be A High-Performance OS Brock McNuggets <brock.mcnuggets@gmail.com> - 2026-11-17 21:52 -0700
    Re: Why macOS Will Never Be A High-Performance OS Jason H <jason_hindle@yahoo.com> - 2026-11-20 21:40 +0000
      Re: Why macOS Will Never Be A High-Performance OS Lawrence D’Oliveiro <ldo@nz.invalid> - 2026-11-20 22:21 +0000
        Re: Why macOS Will Never Be A High-Performance OS Paul <nospam@needed.invalid> - 2026-11-20 19:09 -0500
          Re: Why macOS Will Never Be A High-Performance OS Lawrence D’Oliveiro <ldo@nz.invalid> - 2026-11-21 00:36 +0000
          Re: Why macOS Will Never Be A High-Performance OS ...w¡ñ§±¤ñ <winstonmvp@gmail.com> - 2026-11-20 18:36 -0700
            Re: Why macOS Will Never Be A High-Performance OS Lawrence D’Oliveiro <ldo@nz.invalid> - 2026-11-21 04:40 +0000
              Re: Why macOS Will Never Be A High-Performance OS Paul <nospam@needed.invalid> - 2026-11-21 01:26 -0500
                Re: Why macOS Will Never Be A High-Performance OS Lawrence D’Oliveiro <ldo@nz.invalid> - 2026-11-21 08:19 +0000
                  Re: Why macOS Will Never Be A High-Performance OS Paul <nospam@needed.invalid> - 2026-11-21 06:12 -0500
        Re: Why macOS Will Never Be A High-Performance OS rbowman <bowman@montana.com> - 2026-11-21 01:05 +0000

Page 1 of 2  [1] 2  Next page →


#95569 — Why macOS Will Never Be A High-Performance OS

FromLawrence D’Oliveiro <ldo@nz.invalid>
Date2026-11-12 20:13 +0000
SubjectWhy macOS Will Never Be A High-Performance OS
Message-ID<11d56lf$8idc$2@dont-email.me>
So the latest libdbus update fixes some vulnerabilities
<https://lists.freedesktop.org/archives/dbus/2026-November/018530.html>,
including ones specific to various BSD derivatives. But I couldn’t
quite believe my eyes when I read this:

    On platforms where poll() is emulated using select() (in practice
    this is believed to mean macOS and Interix), prevent a
    denial-of-service in which the system bus would crash if file
    descriptors above 1024 were polled, and harden it against
    busy-looping if an invalid file descriptor was polled.

Really?? The macOS kernel does *not* provide a native poll() call, it
has to emulate it via select()? The whole point of introducing poll()
<https://manpages.debian.org/poll(2)> was to get around the fixed
limit on the number and range of file descriptors
<https://manpages.debian.org/select(2)>, and this is a waste of time
on macOS anyway, because it still has to be implemented via select()??

That is so sad ...

[toc] | [next] | [standalone]


#95580

FromAlan <nuh-uh@nope.com>
Date2026-11-12 14:32 -0700
Message-ID<11d5baf$al3l$2@dont-email.me>
In reply to#95569
On 2026-11-12 13:13, Lawrence D’Oliveiro wrote:
> So the latest libdbus update fixes some vulnerabilities
> <https://lists.freedesktop.org/archives/dbus/2026-November/018530.html>,
> including ones specific to various BSD derivatives. But I couldn’t
> quite believe my eyes when I read this:
> 
>      On platforms where poll() is emulated using select() (in practice
>      this is believed to mean macOS and Interix), prevent a
>      denial-of-service in which the system bus would crash if file
>      descriptors above 1024 were polled, and harden it against
>      busy-looping if an invalid file descriptor was polled.
> 
> Really?? The macOS kernel does *not* provide a native poll() call, it
> has to emulate it via select()? The whole point of introducing poll()
> <https://manpages.debian.org/poll(2)> was to get around the fixed
> limit on the number and range of file descriptors
> <https://manpages.debian.org/select(2)>, and this is a waste of time
> on macOS anyway, because it still has to be implemented via select()??
> 
> That is so sad ...

1. "this is believed to mean macOS". So they don't even know.

2. What practical impact does this have on the ACTUAL USE CASES for macOS?

[toc] | [prev] | [next] | [standalone]


#95599

FromThe Natural Philosopher <tnp@invalid.invalid>
Date2026-11-13 10:35 +0000
Message-ID<11d6p4q$o3fu$2@dont-email.me>
In reply to#95580
On 12/11/2026 21:32, Alan wrote:
> On 2026-11-12 13:13, Lawrence D’Oliveiro wrote:
>> So the latest libdbus update fixes some vulnerabilities
>> <https://lists.freedesktop.org/archives/dbus/2026-November/018530.html>,
>> including ones specific to various BSD derivatives. But I couldn’t
>> quite believe my eyes when I read this:
>>
>>      On platforms where poll() is emulated using select() (in practice
>>      this is believed to mean macOS and Interix), prevent a
>>      denial-of-service in which the system bus would crash if file
>>      descriptors above 1024 were polled, and harden it against
>>      busy-looping if an invalid file descriptor was polled.
>>
>> Really?? The macOS kernel does *not* provide a native poll() call, it
>> has to emulate it via select()? The whole point of introducing poll()
>> <https://manpages.debian.org/poll(2)> was to get around the fixed
>> limit on the number and range of file descriptors
>> <https://manpages.debian.org/select(2)>, and this is a waste of time
>> on macOS anyway, because it still has to be implemented via select()??
>>
>> That is so sad ...
> 
> 1. "this is believed to mean macOS". So they don't even know.
> 
> 2. What practical impact does this have on the ACTUAL USE CASES for macOS?

Did you know there is no proper name for the back of the knee?
-- 
"Nature does not give up the winter because people dislike the cold."

― Confucius

[toc] | [prev] | [next] | [standalone]


#95583

FromCrudeSausage <crude@sausa.ge>
Date2026-11-12 17:15 -0500
Message-ID<6af63a70$1$19$882e4bbb@reader.netnews.com>
In reply to#95569
On 2026-11-12 3:13 p.m., Lawrence D’Oliveiro wrote:
> So the latest libdbus update fixes some vulnerabilities
> <https://lists.freedesktop.org/archives/dbus/2026-November/018530.html>,
> including ones specific to various BSD derivatives. But I couldn’t
> quite believe my eyes when I read this:
> 
>      On platforms where poll() is emulated using select() (in practice
>      this is believed to mean macOS and Interix), prevent a
>      denial-of-service in which the system bus would crash if file
>      descriptors above 1024 were polled, and harden it against
>      busy-looping if an invalid file descriptor was polled.
> 
> Really?? The macOS kernel does *not* provide a native poll() call, it
> has to emulate it via select()? The whole point of introducing poll()
> <https://manpages.debian.org/poll(2)> was to get around the fixed
> limit on the number and range of file descriptors
> <https://manpages.debian.org/select(2)>, and this is a waste of time
> on macOS anyway, because it still has to be implemented via select()??
> 
> That is so sad ...

"Oh no!" said no one at all.

-- 
CrudeSausage
MacBook Air M4 Sky Blue
iPhone 13
Friendly Apple zealot

[toc] | [prev] | [next] | [standalone]


#95609

FromRichard Kettlewell <invalid@invalid.invalid>
Date2026-11-13 19:24 +0000
Message-ID<wwv5wy0cz69.fsf@LkoBDZeT.terraraq.uk>
In reply to#95569
Lawrence D’Oliveiro <ldo@nz.invalid> writes:
> So the latest libdbus update fixes some vulnerabilities
> <https://lists.freedesktop.org/archives/dbus/2026-November/018530.html>,
> including ones specific to various BSD derivatives. But I couldn’t
> quite believe my eyes when I read this:
>
>     On platforms where poll() is emulated using select() (in practice
>     this is believed to mean macOS and Interix), prevent a
>     denial-of-service in which the system bus would crash if file
>     descriptors above 1024 were polled, and harden it against
>     busy-looping if an invalid file descriptor was polled.
>
> Really?? The macOS kernel does *not* provide a native poll() call, it
> has to emulate it via select()?

Why would you post that without checking? It does have poll().

https://github.com/apple-oss-distributions/xnu/blob/main/bsd/kern/sys_generic.c#L1681

> That is so sad ...

Well, something is sad, but I don’t think it’s macOS, in this instance.

-- 
https://www.greenend.org.uk/rjk/

[toc] | [prev] | [next] | [standalone]


#95624

FromRich <rich@example.invalid>
Date2026-11-14 04:02 +0000
Message-ID<11d8mgu$1l22k$1@dont-email.me>
In reply to#95609
In comp.os.linux.misc Richard Kettlewell <invalid@invalid.invalid> wrote:
> Lawrence D’Oliveiro <ldo@nz.invalid> writes:
>> So the latest libdbus update fixes some vulnerabilities 
>> <https://lists.freedesktop.org/archives/dbus/2026-November/018530.html>, 
>> including ones specific to various BSD derivatives.  But I couldn’t 
>> quite believe my eyes when I read this:
>>
>>     On platforms where poll() is emulated using select() (in 
>>     practice this is believed to mean macOS and Interix), prevent a 
>>     denial-of-service in which the system bus would crash if file 
>>     descriptors above 1024 were polled, and harden it against 
>>     busy-looping if an invalid file descriptor was polled.
>>
>> Really?? The macOS kernel does *not* provide a native poll() call, it
>> has to emulate it via select()?
> 
> Why would you post that without checking? It does have poll().
> 
> https://github.com/apple-oss-distributions/xnu/blob/main/bsd/kern/sys_generic.c#L1681

Troll's never check the facts of the things they post that they hope 
will rile some group members up.  The point is to rile folks up, and 
fact checking just gets in the way of that purpose.

>> That is so sad ...
> 
> Well, something is sad, but I don’t think it’s macOS, in this instance.

Indeed.

[toc] | [prev] | [next] | [standalone]


#95634

FromThe Natural Philosopher <tnp@invalid.invalid>
Date2026-11-14 12:07 +0000
Message-ID<11d9it4$1t4e5$10@dont-email.me>
In reply to#95624
On 14/11/2026 04:02, Rich wrote:
> Troll's never check the facts of the things they post that they hope
> will rile some group members up.  The point is to rile folks up, and
> fact checking just gets in the way of that purpose.
> 
Cf  King Donald


-- 
“People believe certain stories because everyone important tells them, 
and people tell those stories because everyone important believes them. 
Indeed, when a conventional wisdom is at its fullest strength, one’s 
agreement with that conventional wisdom becomes almost a litmus test of 
one’s suitability to be taken seriously.”

Paul Krugman

[toc] | [prev] | [next] | [standalone]


#95633

FromThe Natural Philosopher <tnp@invalid.invalid>
Date2026-11-14 12:06 +0000
Message-ID<11d9irj$1t4e5$9@dont-email.me>
In reply to#95609
On 13/11/2026 19:24, Richard Kettlewell wrote:
> Lawrence D’Oliveiro <ldo@nz.invalid> writes:
>> So the latest libdbus update fixes some vulnerabilities
>> <https://lists.freedesktop.org/archives/dbus/2026-November/018530.html>,
>> including ones specific to various BSD derivatives. But I couldn’t
>> quite believe my eyes when I read this:
>>
>>      On platforms where poll() is emulated using select() (in practice
>>      this is believed to mean macOS and Interix), prevent a
>>      denial-of-service in which the system bus would crash if file
>>      descriptors above 1024 were polled, and harden it against
>>      busy-looping if an invalid file descriptor was polled.
>>
>> Really?? The macOS kernel does *not* provide a native poll() call, it
>> has to emulate it via select()?
> 
> Why would you post that without checking? It does have poll().
> 
Its Lawrence, Richard.
#define TRUTH assert(truth)

> https://github.com/apple-oss-distributions/xnu/blob/main/bsd/kern/sys_generic.c#L1681
> 
>> That is so sad ...
> 
> Well, something is sad, but I don’t think it’s macOS, in this instance.
> 
Indeed

-- 
“People believe certain stories because everyone important tells them, 
and people tell those stories because everyone important believes them. 
Indeed, when a conventional wisdom is at its fullest strength, one’s 
agreement with that conventional wisdom becomes almost a litmus test of 
one’s suitability to be taken seriously.”

Paul Krugman

[toc] | [prev] | [next] | [standalone]


#95732

FromWolfFan <akwolffan@zoho.com>
Date2026-11-16 15:43 -0500
Message-ID<0001HW.30ABA254026EF0157000014BE38F@news.supernews.com>
In reply to#95609
On Nov 13, 2026, Richard Kettlewell wrote
(in article <wwv5wy0cz69.fsf@LkoBDZeT.terraraq.uk>):

> Lawrence D’Oliveiro <ldo@nz.invalid> writes:
> > So the latest libdbus update fixes some vulnerabilities
> > <https://lists.freedesktop.org/archives/dbus/2026-November/018530.html>,
> > including ones specific to various BSD derivatives. But I couldn’t
> > quite believe my eyes when I read this:
> >
> > On platforms where poll() is emulated using select() (in practice
> > this is believed to mean macOS and Interix), prevent a
> > denial-of-service in which the system bus would crash if file
> > descriptors above 1024 were polled, and harden it against
> > busy-looping if an invalid file descriptor was polled.
> >
> > Really?? The macOS kernel does *not* provide a native poll() call, it
> > has to emulate it via select()?
>
> Why would you post that without checking?

because he’s a know-nothing troll.
> It does have poll().
>
> https://github.com/apple-oss-distributions/xnu/blob/main/bsd/kern/sys_generic.
> c#L1681
>
> > That is so sad ...
>
> Well, something is sad, but I don’t think it’s macOS, in this instance.

Anti-Apple trolls have been posting falsehoods for decades. He’s not even 
original.

[toc] | [prev] | [next] | [standalone]


#95733

FromNick Charles <none@none.none>
Date2026-11-16 21:21 +0000
Message-ID<51WdnQeST_FB7mb3nZ2dnZfqnPWdnZ2d@supernews.com>
In reply to#95732
On Nov 16, 2026 at 3:43:00 PM EST, "WolfFan" <akwolffan@zoho.com> wrote:

> On Nov 13, 2026, Richard Kettlewell wrote
> (in article <wwv5wy0cz69.fsf@LkoBDZeT.terraraq.uk>):
> 
>> Lawrence D’Oliveiro <ldo@nz.invalid> writes:
>>> So the latest libdbus update fixes some vulnerabilities
>>> <https://lists.freedesktop.org/archives/dbus/2026-November/018530.html>,
>>> including ones specific to various BSD derivatives. But I couldn’t
>>> quite believe my eyes when I read this:
>>> 
>>> On platforms where poll() is emulated using select() (in practice
>>> this is believed to mean macOS and Interix), prevent a
>>> denial-of-service in which the system bus would crash if file
>>> descriptors above 1024 were polled, and harden it against
>>> busy-looping if an invalid file descriptor was polled.
>>> 
>>> Really?? The macOS kernel does *not* provide a native poll() call, it
>>> has to emulate it via select()?
>> 
>> Why would you post that without checking?
> 
> because he’s a know-nothing troll.

Some would say "know-nothing troll" is redundant.

[toc] | [prev] | [next] | [standalone]


#95749

FromLawrence D’Oliveiro <ldo@nz.invalid>
Date2026-11-18 04:43 +0000
Message-ID<11djae1$11uc6$1@dont-email.me>
In reply to#95569
Does make you wonder what they were running on those Icloud servers
that have fallen over. Was it old XServe boxes that they couldn’t
sell?

[toc] | [prev] | [next] | [standalone]


#95750

FromBrock McNuggets <brock.mcnuggets@gmail.com>
Date2026-11-17 21:52 -0700
Message-ID<1794977573.BB3AC18A-7476-43A9-9154-1F26040477D3@gmail.com>
In reply to#95569
On 12 Nov 2026, Lawrence D´Oliveiro wrote in
<11d56lf$8idc$2@dont-email.me>:
> So the latest libdbus update fixes some vulnerabilities
> <https://lists.freedesktop.org/archives/dbus/2026-November/018530.html>,
> including ones specific to various BSD derivatives. But I couldn't
> quite believe my eyes when I read this:
> 
>     On platforms where poll() is emulated using select() (in practice
>     this is believed to mean macOS and Interix), prevent a
>     denial-of-service in which the system bus would crash if file
>     descriptors above 1024 were polled, and harden it against
>     busy-looping if an invalid file descriptor was polled.
> 
> Really?? The macOS kernel does *not* provide a native poll() call,
> it has to emulate it via select()? The whole point of introducing
> poll() <https://manpages.debian.org/poll(2)> was to get around the
> fixed limit on the number and range of file descriptors
> <https://manpages.debian.org/select(2)>, and this is a waste of
> time on macOS anyway, because it still has to be implemented via
> select()??
> 
> That is so sad ...
> 

Look into kqueue. 

-- 
You can't find peace with someone at war with themselves.

[toc] | [prev] | [next] | [standalone]


#95807

FromJason H <jason_hindle@yahoo.com>
Date2026-11-20 21:40 +0000
Message-ID<11dqeo7$3insq$1@dont-email.me>
In reply to#95569
On 12/11/2026 20:13, Lawrence DOliveiro wrote:
>So the latest libdbus update fixes some vulnerabilities
><https://lists.freedesktop.org/archives/dbus/2026-November/018530.html>,
>including ones specific to various BSD derivatives. But I couldn’t
>quite believe my eyes when I read this:
>
>    On platforms where poll() is emulated using select() (in practice
>    this is believed to mean macOS and Interix), prevent a
>    denial-of-service in which the system bus would crash if file
>    descriptors above 1024 were polled, and harden it against
>    busy-looping if an invalid file descriptor was polled.
>
>Really?? The macOS kernel does *not* provide a native poll() call, it
>has to emulate it via select()? The whole point of introducing poll()
><https://manpages.debian.org/poll(2)> was to get around the fixed
>limit on the number and range of file descriptors
><https://manpages.debian.org/select(2)>, and this is a waste of time
>on macOS anyway, because it still has to be implemented via select()??
>
>That is so sad ...

Apple ceded server computing, where this is far more likely to become an
 issue, to Linux. They're simply not in that space anymore. As business
 decisions go, it was a good one, and one that Microsoft can't quite bring
 itself to make...

-- 
--
A PICKER OF UNCONSIDERED TRIFLES

[toc] | [prev] | [next] | [standalone]


#95808

FromLawrence D’Oliveiro <ldo@nz.invalid>
Date2026-11-20 22:21 +0000
Message-ID<11dqh5l$3j82r$2@dont-email.me>
In reply to#95807
On Fri, 20 Nov 2026 21:40:23 -0000 (UTC), Jason H wrote:

> Apple ceded server computing, where this is far more likely to
> become an issue, to Linux. They're simply not in that space anymore.
> As business decisions go, it was a good one, and one that Microsoft
> can't quite bring itself to make...

Actually, Microsoft *is* doing the same sort of thing. Its cloud
service is predominantly Linux-based, and revenue from that is already
several times that coming from the division that includes Windows and
Xbox and a few other lesser products and services.

If you look at the development of Windows Server, you can see the
on-prem version already taking a back seat to the cloud version. And
the cloud version is in turn completely outnumbered by the customers
using Linux.

Wonder why the quality of Windows updates seems to be deteriorating
lately? The simple fact is, the revenue coming from that is in
decline, and so investment back into that business declines with it.

[toc] | [prev] | [next] | [standalone]


#95809

FromPaul <nospam@needed.invalid>
Date2026-11-20 19:09 -0500
Message-ID<11dqngb$3lnpu$1@dont-email.me>
In reply to#95808
On Fri, 11/20/2026 5:21 PM, Lawrence D’Oliveiro wrote:
> On Fri, 20 Nov 2026 21:40:23 -0000 (UTC), Jason H wrote:
> 
>> Apple ceded server computing, where this is far more likely to
>> become an issue, to Linux. They're simply not in that space anymore.
>> As business decisions go, it was a good one, and one that Microsoft
>> can't quite bring itself to make...
> 
> Actually, Microsoft *is* doing the same sort of thing. Its cloud
> service is predominantly Linux-based, and revenue from that is already
> several times that coming from the division that includes Windows and
> Xbox and a few other lesser products and services.
> 
> If you look at the development of Windows Server, you can see the
> on-prem version already taking a back seat to the cloud version. And
> the cloud version is in turn completely outnumbered by the customers
> using Linux.
> 
> Wonder why the quality of Windows updates seems to be deteriorating
> lately? The simple fact is, the revenue coming from that is in
> decline, and so investment back into that business declines with it.

Um, the Windows Updates are currently augmented by Mythos or Fable
scans for exploits. The resulting 1000 or 1400 security patches,
are bound to have some impact on operations. This is a "hurry-up"
activity, before the Black Hats do their fuzzing and scans with
Mythos or Fable.

How the Windows Group operates has nothing to do with revenue
in a direct one-to-one relationship. It operates on the promise
that some day, it will be a larger source of income. It doesn't
exist to sell copies on DVD, it exists to sell Microsoft Office,
and soon, to rent tokens.

Remember that for the longest while, a huge pile of cash
insulated the operation from outside influence. Now, by
betting the farm on datacenters and other chattels, they
no longer have the luxury of ignoring things they once ignored.
They're following the lead of Oracle, getting over the tips
of their skis and so on. The investments they're making,
remove protection of internal operations. This is why we
have things happening like 60,000 packages offered to employees
to leave.

They are all-in on LLM-AI, and *finally* they have a subscription
model with legs. Previously the idea of subscriptions was weak.
Now they have a tangible thing to sell -- tokens. The part that
remains to be seen, is the "reveal" of just what the tokens will be
doing. Writing a ransom note ? We can help with that :-) "Recall"
is slightly different, in that a portion of it runs locally, and
on a 50 TOPS NPU present in a limited set of CPUs. There was
a recent announcement of "plans" and what they were for.

The Windows Desktop OS continues to exist as a sales vehicle. It's
a platform for experiments. And for subscription income. When they
can come up with a compelling reason for a subscription.

   Paul

[toc] | [prev] | [next] | [standalone]


#95810

FromLawrence D’Oliveiro <ldo@nz.invalid>
Date2026-11-21 00:36 +0000
Message-ID<11dqp2o$3m4hd$1@dont-email.me>
In reply to#95809
On Fri, 20 Nov 2026 19:09:48 -0500, Paul wrote:

> On Fri, 11/20/2026 5:21 PM, Lawrence D’Oliveiro wrote:
>>
>> Wonder why the quality of Windows updates seems to be deteriorating
>> lately? The simple fact is, the revenue coming from that is in
>> decline, and so investment back into that business declines with
>> it.
>
> Um, the Windows Updates are currently augmented by Mythos or Fable
> scans for exploits. The resulting 1000 or 1400 security patches,
> are bound to have some impact on operations. This is a "hurry-up"
> activity, before the Black Hats do their fuzzing and scans with
> Mythos or Fable.

How is it that other groups, like the Linux developers, can fix this
sort of thing without introducing new problems? Yet all too often
these days, Microsoft has to rush out an “emergency update” to fix
problems caused by their last update.

> How the Windows Group operates has nothing to do with revenue
> in a direct one-to-one relationship. It operates on the promise
> that some day, it will be a larger source of income.

Except there is no such promise any more. We are long past the point
of “peak Windows”.

> The Windows Desktop OS continues to exist as a sales vehicle.

It doesn’t seem to be selling much these days. If you want to do
development on it, one of the first things you do is install WSL so
you can run Linux.

[toc] | [prev] | [next] | [standalone]


#95812

From...w¡ñ§±¤ñ <winstonmvp@gmail.com>
Date2026-11-20 18:36 -0700
Message-ID<11dqsjm$3n6cl$1@dont-email.me>
In reply to#95809
On 11/20/26 5:09 PM, Paul wrote:

> 
> The Windows Desktop OS continues to exist as a sales vehicle. It's
> a platform for experiments. And for subscription income. When they
> can come up with a compelling reason for a subscription.
> 
>     Paul

Microsoft already has a subscription version of Windows(E3/E5).
  In 2025 increased by 15%, though the exact quantity is not reported as 
a stand-alone unit.

M365(not consumer) Commercial seats was over 500 million in 2026 through 
March 2026 which, in Enterprise includes a subset of subscription 
Windows seats(per seat, volume licensing).

The future may very well be a subscription version for consumers, but 
that will in most part be determined on the OEM pc builders contractual 
preference or agreed upon MSFT and OEM policy.

i.e. the compelling reason is and always has been(for Windows 
subscription for consumers) a moving target.

For Enterprise - the same applies to a certain extent because Enterprise 
purchase OEM devices with Windows(often at a discount related to 
volume), archive the existing OEM license for internal policy 
requirments, then deploy volume licensing across the organization.

-- 
...w¡ñ§±¤ñ

[toc] | [prev] | [next] | [standalone]


#95818

FromLawrence D’Oliveiro <ldo@nz.invalid>
Date2026-11-21 04:40 +0000
Message-ID<11dr7b3$3q0pa$1@dont-email.me>
In reply to#95812
On Fri, 20 Nov 2026 18:36:53 -0700, ...w¡ñ§±¤ñ wrote:

> Microsoft already has a subscription version of Windows(E3/E5).
>   In 2025 increased by 15%, though the exact quantity is not reported as 
> a stand-alone unit.

Microsoft is getting 83% of its revenue from cloud services, and only
17% from the division that includes Microsoft Windows
<https://www.theverge.com/tech/1020785/microsoft-earnings-revenue-q1-2027>:

    Agents and Infra, which includes businesses like Azure, Microsoft
    365 commercial and consumer, and server products, makes up 83
    percent ($76.8 billion) of the overall $92.3 billion of
    Microsoft’s revenue and has increased 25 percent year-over-year.
    The big changes to Microsoft’s financial reporting also mean we
    will now see the company disclose Azure revenue on a quarterly
    basis. Azure contributed $33.17 billion in revenue this quarter,
    up 48 percent year-over-year.

    Devices and Consumer, which includes Windows, Xbox, Surface, and
    Microsoft’s search and advertising businesses, makes up the
    remaining 17 percent ($15.5 billion) of Microsoft’s $92.3 billion
    in revenues this quarter — a decrease of three percent.

But those cloud services are predominantly running on Linux servers.
Which means that Linux has become far more important to Microsoft as
the foundation for its entire business than its own Windows OS.

Wonder why the quality of Windows updates continues to decline? Why
patches often don’t fix problems properly, indeed they often create
new problems? Because the whole Windows business simply isn’t
profitable enough to justify much ongoing investment any more. And
it’s only going to get worse, not better:

    Microsoft CFO Amy Hood said on the company’s earnings call that
    she expects Windows OEM revenue to decline again next quarter,
    around “the mid 30s,” and a decline is expected to continue
    throughout the 2027 financial year.

[toc] | [prev] | [next] | [standalone]


#95819

FromPaul <nospam@needed.invalid>
Date2026-11-21 01:26 -0500
Message-ID<11drdhr$3rlb7$1@dont-email.me>
In reply to#95818
On Fri, 11/20/2026 11:40 PM, Lawrence D’Oliveiro wrote:

> Wonder why the quality of Windows updates continues to decline? Why
> patches often don’t fix problems properly, indeed they often create
> new problems? Because the whole Windows business simply isn’t
> profitable enough to justify much ongoing investment any more. And
> it’s only going to get worse, not better:
> 
>     Microsoft CFO Amy Hood said on the company’s earnings call that
>     she expects Windows OEM revenue to decline again next quarter,
>     around “the mid 30s,” and a decline is expected to continue
>     throughout the 2027 financial year.

Intel's sales were off by about 22% or so.

It's not as if one company stands alone in being
affected by the Tulip Craze.

This is why we're making jokes about 8GB machines.
That's what they want to ship now.

*******

Patching is just as complicated in Windows, as in Linux.
Ubuntu has SNAP updates, async to other patching. Windows
has App updates (I got ten of them today), async to Patch Tuesday.
Windows has multiple streams, and things need just as much
backpatching as Linux does. Windows has at least a thousand
packages that make up the OS.

My Windows side-by-side folder

   WinSxS

   Size         16.4GB
   Size on disk  9.81GB <=== and hard linked to "where they are used" on disk
   124,061 files

A lot of those files are tiny, and live in the $MFT. But they're
all part of bookkeeping for the repository.

One App level package, can receive nine updates in one day.
Why is that ? Well, that's a labeling problem. There is no
room in the label, to indicate they're actually different versions
of the same thing. And all nine of them are getting patched.
And sometimes, they all get patched within the same minute!

It's rather busy in there.

   Paul

[toc] | [prev] | [next] | [standalone]


#95821

FromLawrence D’Oliveiro <ldo@nz.invalid>
Date2026-11-21 08:19 +0000
Message-ID<11drk6b$3tcnm$1@dont-email.me>
In reply to#95819
On Sat, 21 Nov 2026 01:26:01 -0500, Paul wrote:

> On Fri, 11/20/2026 11:40 PM, Lawrence D’Oliveiro wrote:
>
>> Wonder why the quality of Windows updates continues to decline? Why
>> patches often don’t fix problems properly, indeed they often create
>> new problems? Because the whole Windows business simply isn’t
>> profitable enough to justify much ongoing investment any more. And
>> it’s only going to get worse, not better:
>>
>>     Microsoft CFO Amy Hood said on the company’s earnings call that
>>     she expects Windows OEM revenue to decline again next quarter,
>>     around “the mid 30s,” and a decline is expected to continue
>>     throughout the 2027 financial year.
>
> Intel's sales were off by about 22% or so.
>
> It's not as if one company stands alone in being affected by the
> Tulip Craze.

But Microsoft’s cloud business is booming. It’s the Windows business
(and Xbox) that are not doing so well. The company as a whole is not
hurting at all.

> Patching is just as complicated in Windows, as in Linux. Ubuntu has
> SNAP updates, async to other patching. Windows has App updates (I
> got ten of them today), async to Patch Tuesday. Windows has multiple
> streams, and things need just as much backpatching as Linux does.
> Windows has at least a thousand packages that make up the OS.

Other Linux distros are not as enamoured of snaps as Ubuntu is. And
they pretty much all have package management systems that can scale
gracefully to updating thousands of packages at once.

And then you have the issue that Windows insists on “locking” files
that are currently open by running processes, which is why system
updates almost always render the system unusable, and require a reboot
afterwards.

[toc] | [prev] | [next] | [standalone]


Page 1 of 2  [1] 2  Next page →

Back to top | Article view | comp.os.linux.misc


csiph-web