• Unix is dead

    From Ben Collver@21:1/5 to All on Tue Jan 17 18:15:12 2023
    Unix is dead. Long live Unix!

    Don't expect to see any more big AIX news. This means the last Unix
    left is... Linux

    Liam Proven
    Tue 17 Jan 2023

    It's the end of an era. As The Reg covered last week, IBM has
    transferred development of AIX to India. Why should IBM pay for an
    expensive US-based team to maintain its own proprietary flavor of
    official Unix when it paid 34 billion bucks for its own FOSS flavor
    in Red Hat?

    Here at The Reg FOSS desk, we've felt this was coming ever since we
    reported that Big Blue was launching new POWER servers which didn't
    support AIX--already nearly eight years ago. Even if it was visibly
    coming over the horizon, this is a significant event: AIX is the last proprietary Unix which was in active development, and constitutes
    four of the 10 entries in the official Open Group list.

    Within Oracle, Solaris is in maintenance mode. Almost exactly six
    year ago, we reported that the next major release, Solaris 12, had
    disappeared from Oracle's roadmap. HPE's HP-UX is also in
    maintenance mode because there's no new hardware to run it on.
    Itanium really is dead now and at the end that's all HP-UX could run
    on. It's over a decade since we reported that HP investigated but
    canceled an effort to port it to x86-64.

    The last incarnation of the SCO Group, Xinuos, is still around and
    offers not one but two proprietary UNIX variants: SCO OpenServer,
    descended from SCO Xenix, and UnixWare, descended from Novell's Unix.
    We note that OpenServer 10, a more modern OS based on FreeBSD 10,
    has disappeared from Xinuos's homepage. It's worth pointing out that
    the SCO Group was the company formerly known as Caldera, and isn't
    the same SCO as the Santa Cruz Operation which co-created Xenix with
    Microsoft in the 1980s.

    There used to be two Chinese Linux distros which had passed the Open
    Group's testing and could use the Unix trademark: Inspur K/UX and
    Huawei EulerOS. Both companies have let the rather expensive
    trademark lapse, though. But the important detail here is that Linux
    passed and was certified as a UNIX (tm). And it wasn't just one
    distro, although both were CentOS Linux derivatives. We suspect that
    any Linux would breeze through because several many un-Unix-like OSes
    have passed before.

    Other OSes have passed or probably easily would, though. IBM's z/OS
    is alive and well: version 2.5 came out in 2021 and in 2022 Big Blue
    started offering cloud instances. z/OS has a Unix-compatible
    environment which has passed the compatibility tests so officially,
    it's a UNIX (tm), even if that wasn't its original native API.

    The "open" in the name "OpenVMS" originally referred to the POSIX
    compatibility it gained with version 5, way back in 1991, and was
    first applied to the new version for DEC's Alpha CPUs. Last year VMS
    Software released version 9.2 for x86-64 hypervisors (and a single
    supported box, HPE's DL380).

    Ever since Windows NT in 1993, Windows has had a POSIX environment.
    Now, with WSL, it arguably has two of them, and we suspect that if
    Microsoft were so inclined, it could have Windows certified as an
    official Unix-compatible OS.

    In our recent story on Beta 4 of Haiku, we said it wasn't really a
    Unix. As you can see, there's an editor's note attached to the end
    of the story explaining why.

    We had heard from Haiku's primary full-time developer, who vigorously
    disagreed with our point of view. To his mind, the fact that Haiku
    now has strong Unix compatibility, with some of the main Unix
    directories present in its filesystem, a quite complete set of Unix
    API calls, a Unix shell, and so on, means that Haiku is quite
    definitely a Unix. We feel that inasmuch as it's a reimplementation
    of BeOS, with its own native filesystem, API, GUI and so on, it's
    something different, which offers Unix compatibility as well.

    But this illustrates the difficulty of defining precisely what the
    word "Unix" means in the 21st century. It hasn't meant "based on
    AT&T code" since Novell bought Unix System Labs from AT&T in 1993,
    kept the code, and donated the trademark to the Open Group. Since
    that time, if it passes the Open Group's testing (and you pay a fee
    to use the trademark), it's UNIX (tm). Haiku hasn't so it isn't.
    Linux has so it is. But then so is z/OS, which is a direct
    descendant of OS/390, or IBM MVS as it was called when it was
    launched in 1974. In other words, an OS which isn't actually based
    on, similar to, or even related to Unix.

    Which means that the last officially trademarked commercial UNIX (tm)
    is Apple's macOS 13, which underneath the proprietary GUI layer is
    mostly an open source OS called Darwin anyway. The kernel, XNU, is
    based on Mach with an in-kernel "Unix server" derived from FreeBSD.

    So, as of 2023, open source really has won. There are more Unix-like
    OSes than ever, and some very un-Unix-like OSes which are highly
    compatible with it, but the official line is, to all intents and
    purposes, dead and gone. All the proprietary, commercial Unixes are
    now on life support: they will get essential bug fixes and security
    updates, but we won't be seeing any major new releases.

    Send flowers.

    From: https://www.theregister.com/2023/01/17/unix_is_dead/

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Bob Eager@21:1/5 to Ben Collver on Tue Jan 17 20:13:12 2023
    On Tue, 17 Jan 2023 18:15:12 +0000, Ben Collver wrote:

    Unix is dead. Long live Unix!

    Don't expect to see any more big AIX news. This means the last Unix
    left is... Linux

    macOS is a certified UNIX.

    And Linux isn't.



    --
    Using UNIX since v6 (1975)...

    Use the BIG mirror service in the UK:
    http://www.mirrorservice.org

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Marco Moock@21:1/5 to All on Tue Jan 17 21:39:42 2023
    Am 17.01.2023 um 20:13:12 Uhr schrieb Bob Eager:

    On Tue, 17 Jan 2023 18:15:12 +0000, Ben Collver wrote:

    Unix is dead. Long live Unix!

    Don't expect to see any more big AIX news. This means the last Unix
    left is... Linux

    macOS is a certified UNIX.

    True, but it doesn't have all the benefits of a Linux distribution, but
    all disadvantages from UNIX. It officially runs only on special
    hardware, it lacks things like apt/dnf...

    And Linux isn't.

    Who cares?
    GNU/Linux took over UNIX in the last years. Is there anybody here who
    set up a new UNIX system in the last years?

    What was the reason?

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Marco Moock@21:1/5 to All on Tue Jan 17 21:37:59 2023
    Am 17.01.2023 um 18:15:12 Uhr schrieb Ben Collver:

    It's the end of an era. As The Reg covered last week, IBM has
    transferred development of AIX to India. Why should IBM pay for an
    expensive US-based team to maintain its own proprietary flavor of
    official Unix when it paid 34 billion bucks for its own FOSS flavor
    in Red Hat?

    For me it seems that IBM never wanted AIX to grow - they only support
    that on special hardware. Most Linux distributions run on most hardware
    - and are free of charge.

    Within Oracle, Solaris is in maintenance mode. Almost exactly six
    year ago, we reported that the next major release, Solaris 12, had disappeared from Oracle's roadmap.

    Solaris 11.4 has been released in 2018, in 2017 they laid off many
    employees developing it.

    HPE's HP-UX is also in maintenance mode because there's no new
    hardware to run it on. Itanium really is dead now and at the end
    that's all HP-UX could run on. It's over a decade since we reported
    that HP investigated but canceled an effort to port it to x86-64.

    It seems they want it to die.
    They don't want people to download it without creating an Oracle
    account.

    I have never used a real UNIX - I never managed to get one.
    Maybe the reason is that the time of UNIX is over - I started with
    computers in ~2010 and with Linux in 2015 (in the age of 14).

    We still have some AIX machines at work, but old-ass versions and we
    want to switch them off due to security reasons.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Sylvia Else@21:1/5 to Marco Moock on Wed Jan 18 11:12:03 2023
    On 18-Jan-23 7:37 am, Marco Moock wrote:
    Am 17.01.2023 um 18:15:12 Uhr schrieb Ben Collver:

    It's the end of an era. As The Reg covered last week, IBM has
    transferred development of AIX to India. Why should IBM pay for an
    expensive US-based team to maintain its own proprietary flavor of
    official Unix when it paid 34 billion bucks for its own FOSS flavor
    in Red Hat?

    For me it seems that IBM never wanted AIX to grow - they only support
    that on special hardware. Most Linux distributions run on most hardware
    - and are free of charge.

    Within Oracle, Solaris is in maintenance mode. Almost exactly six
    year ago, we reported that the next major release, Solaris 12, had
    disappeared from Oracle's roadmap.

    Solaris 11.4 has been released in 2018, in 2017 they laid off many
    employees developing it.

    HPE's HP-UX is also in maintenance mode because there's no new
    hardware to run it on. Itanium really is dead now and at the end
    that's all HP-UX could run on. It's over a decade since we reported
    that HP investigated but canceled an effort to port it to x86-64.

    It seems they want it to die.
    They don't want people to download it without creating an Oracle
    account.

    I have never used a real UNIX - I never managed to get one.
    Maybe the reason is that the time of UNIX is over - I started with
    computers in ~2010 and with Linux in 2015 (in the age of 14).

    We still have some AIX machines at work, but old-ass versions and we
    want to switch them off due to security reasons.


    I remember flying from Sydney to Brisbane[*] for a day to investigate a
    problem with some software that wasn't working under the client's
    Solaris system, despite working under the versions of Unix that we
    actually had. Turned out to be a bug in Solaris. I can't remember the
    details.

    If it hadn't been a proprietary operating system, it would have been
    much easier to deal with.

    [*] Two nearby cities on the Australian east coast that are actually
    almost 500 miles apart.

    Sylvia.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Blue-Maned_Hawk@21:1/5 to All on Wed Jan 18 04:21:09 2023
    This is an OpenPGP/MIME signed message (RFC 4880 and 3156) --------------Fy7QA7CCdLW8MCSPI70cRgjF
    Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: base64

    T24gMS8xNy8yMyAxNToxMywgQm9iIEVhZ2VyIHdyb3RlOg0KPiBPbiBUdWUsIDE3IEphbiAy MDIzIDE4OjE1OjEyICswMDAwLCBCZW4gQ29sbHZlciB3cm90ZToNCj4gDQo+PiBVbml4IGlz IGRlYWQuICBMb25nIGxpdmUgVW5peCENCj4+DQo+PiBEb24ndCBleHBlY3QgdG8gc2VlIGFu eSBtb3JlIGJpZyBBSVggbmV3cy4gIFRoaXMgbWVhbnMgdGhlIGxhc3QgVW5peA0KPj4gbGVm dCBpcy4uLiAgTGludXgNCj4gDQo+IG1hY09TIGlzIGEgY2VydGlmaWVkIFVOSVguDQo+IA0K PiBBbmQgTGludXggaXNuJ3QuDQo+IA0KDQrigItJdCBzZWVtcyB0byBtZSBsaWtlIHRoZXJl IG1pZ2h0IGJlIHR3byBkaWZmZXJlbnQgZGVmaW5pdGlvbnMgb2YgIlVuaXgiIA0KYXQgd29y ayBoZXJlLg0KDQotLSANCuKal++4jiB8IC9ibHUubcmbaW4uZMqwYWsvIHwgc2hvcnRlbnMg dG8gIkhhd2siIHwgaGUvaGltL2hpcy9oaW1zZWxmL01yLg0KYmx1ZW1hbmVkaGF3ay5naXRo dWIuaW8NCkFwb2xvZ2llcyBpZiB0aGlzIG1lc3NhZ2UgZ290IGVuY29kZWQgaW50byBvY3Rv Y3RhbC4gIEJsYW1lIFRodW5kZXJiaXJkIA0KZm9yIG5vdCBiZWhhdmluZyBjb3JyZWN0bHkg d2l0aCBVbmljb2RlIGFuZCBQR1AuDQoNCg==

    --------------Fy7QA7CCdLW8MCSPI70cRgjF--

    -----BEGIN PGP SIGNATURE-----

    wnsEABYIACMWIQRyCHAqodVWYEd5ANANjGnZxCulyAUCY8e6BQUDAAAAAAAKCRANjGnZxCulyFiY APwPng8f6+4wZgf7QRuF1s0kQhqNnYz3RA8u/SkL7GKPzQEArGyc5DwFCIxlQZGW+2G3NZiExkX5 rZknKfzaEikH2g0=
    =UPST
    -----END PGP SIGNATURE-----

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Marco Moock@21:1/5 to All on Wed Jan 18 10:58:09 2023
    Am 18.01.2023 um 04:21:09 Uhr schrieb Blue-Maned_Hawk:

    ​It seems to me like there might be two different definitions of
    "Unix" at work here.

    There is UNIX and UNIX-like. Some people confuse them.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Geoff Clare@21:1/5 to Bob Eager on Wed Jan 18 13:38:21 2023
    Bob Eager wrote:

    On Tue, 17 Jan 2023 18:15:12 +0000, Ben Collver wrote:

    Unix is dead. Long live Unix!

    Don't expect to see any more big AIX news. This means the last Unix
    left is... Linux

    macOS is a certified UNIX.

    And Linux isn't.

    Seems you may have missed this part of the article...

    There used to be two Chinese Linux distros which had passed the
    Open Group's testing and could use the Unix trademark: Inspur K/UX
    and Huawei EulerOS. Both companies have let the rather expensive
    trademark lapse, though. But the important detail here is that Linux
    passed and was certified as a UNIX (tm).

    (I say "may" because, by using the present tense, you were technically
    correct, but don't think that was the point of your post.)

    --
    Geoff Clare <netnews@gclare.org.uk>

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Grant Taylor@21:1/5 to Marco Moock on Wed Jan 18 22:12:17 2023
    On 1/18/23 2:58 AM, Marco Moock wrote:
    There is UNIX and UNIX-like. Some people confuse them.

    I usually refer to these as Unix and unix (like OS).

    The capital "U" is important.



    --
    Grant. . . .
    unix || die

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Y A@21:1/5 to Ben Collver on Sat Feb 11 08:07:21 2023
    As far as the linux lives, unix will never die. I am using pop linux.




    On Tuesday, January 17, 2023 at 8:15:21 PM UTC+2, Ben Collver wrote:
    Unix is dead. Long live Unix!

    Don't expect to see any more big AIX news. This means the last Unix
    left is... Linux

    Liam Proven
    Tue 17 Jan 2023

    It's the end of an era. As The Reg covered last week, IBM has
    transferred development of AIX to India. Why should IBM pay for an
    expensive US-based team to maintain its own proprietary flavor of
    official Unix when it paid 34 billion bucks for its own FOSS flavor
    in Red Hat?

    Here at The Reg FOSS desk, we've felt this was coming ever since we
    reported that Big Blue was launching new POWER servers which didn't
    support AIX--already nearly eight years ago. Even if it was visibly
    coming over the horizon, this is a significant event: AIX is the last proprietary Unix which was in active development, and constitutes
    four of the 10 entries in the official Open Group list.

    Within Oracle, Solaris is in maintenance mode. Almost exactly six
    year ago, we reported that the next major release, Solaris 12, had disappeared from Oracle's roadmap. HPE's HP-UX is also in
    maintenance mode because there's no new hardware to run it on.
    Itanium really is dead now and at the end that's all HP-UX could run
    on. It's over a decade since we reported that HP investigated but
    canceled an effort to port it to x86-64.

    The last incarnation of the SCO Group, Xinuos, is still around and
    offers not one but two proprietary UNIX variants: SCO OpenServer,
    descended from SCO Xenix, and UnixWare, descended from Novell's Unix.
    We note that OpenServer 10, a more modern OS based on FreeBSD 10,
    has disappeared from Xinuos's homepage. It's worth pointing out that
    the SCO Group was the company formerly known as Caldera, and isn't
    the same SCO as the Santa Cruz Operation which co-created Xenix with Microsoft in the 1980s.

    There used to be two Chinese Linux distros which had passed the Open
    Group's testing and could use the Unix trademark: Inspur K/UX and
    Huawei EulerOS. Both companies have let the rather expensive
    trademark lapse, though. But the important detail here is that Linux
    passed and was certified as a UNIX (tm). And it wasn't just one
    distro, although both were CentOS Linux derivatives. We suspect that
    any Linux would breeze through because several many un-Unix-like OSes
    have passed before.

    Other OSes have passed or probably easily would, though. IBM's z/OS
    is alive and well: version 2.5 came out in 2021 and in 2022 Big Blue
    started offering cloud instances. z/OS has a Unix-compatible
    environment which has passed the compatibility tests so officially,
    it's a UNIX (tm), even if that wasn't its original native API.

    The "open" in the name "OpenVMS" originally referred to the POSIX compatibility it gained with version 5, way back in 1991, and was
    first applied to the new version for DEC's Alpha CPUs. Last year VMS
    Software released version 9.2 for x86-64 hypervisors (and a single
    supported box, HPE's DL380).

    Ever since Windows NT in 1993, Windows has had a POSIX environment.
    Now, with WSL, it arguably has two of them, and we suspect that if
    Microsoft were so inclined, it could have Windows certified as an
    official Unix-compatible OS.

    In our recent story on Beta 4 of Haiku, we said it wasn't really a
    Unix. As you can see, there's an editor's note attached to the end
    of the story explaining why.

    We had heard from Haiku's primary full-time developer, who vigorously disagreed with our point of view. To his mind, the fact that Haiku
    now has strong Unix compatibility, with some of the main Unix
    directories present in its filesystem, a quite complete set of Unix
    API calls, a Unix shell, and so on, means that Haiku is quite
    definitely a Unix. We feel that inasmuch as it's a reimplementation
    of BeOS, with its own native filesystem, API, GUI and so on, it's
    something different, which offers Unix compatibility as well.

    But this illustrates the difficulty of defining precisely what the
    word "Unix" means in the 21st century. It hasn't meant "based on
    AT&T code" since Novell bought Unix System Labs from AT&T in 1993,
    kept the code, and donated the trademark to the Open Group. Since
    that time, if it passes the Open Group's testing (and you pay a fee
    to use the trademark), it's UNIX (tm). Haiku hasn't so it isn't.
    Linux has so it is. But then so is z/OS, which is a direct
    descendant of OS/390, or IBM MVS as it was called when it was
    launched in 1974. In other words, an OS which isn't actually based
    on, similar to, or even related to Unix.

    Which means that the last officially trademarked commercial UNIX (tm)
    is Apple's macOS 13, which underneath the proprietary GUI layer is
    mostly an open source OS called Darwin anyway. The kernel, XNU, is
    based on Mach with an in-kernel "Unix server" derived from FreeBSD.

    So, as of 2023, open source really has won. There are more Unix-like
    OSes than ever, and some very un-Unix-like OSes which are highly
    compatible with it, but the official line is, to all intents and
    purposes, dead and gone. All the proprietary, commercial Unixes are
    now on life support: they will get essential bug fixes and security
    updates, but we won't be seeing any major new releases.

    Send flowers.

    From: https://www.theregister.com/2023/01/17/unix_is_dead/

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Ian McCall@21:1/5 to All on Thu Feb 16 09:19:00 2023
    On 11 Feb 2023, Y A wrote
    (in article<420c786f-934f-4bbc-b4d2-91793af96655n@googlegroups.com>):

    As far as the linux lives, unix will never die. I am using pop linux.

    On Tuesday, January 17, 2023 at 8:15:21 PM UTC+2, Ben Collver wrote:
    Unix is dead. Long live Unix!

    Don't expect to see any more big AIX news. This means the last Unix
    left is... Linux

    Liam Proven
    Tue 17 Jan 2023

    Linux isn’t Unix. They appear to have missed off the most commercially successful Unix...macOS.

    Cheers,
    Ian

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Grant Taylor@21:1/5 to Ian McCall on Thu Feb 16 10:21:09 2023
    On 2/16/23 2:19 AM, Ian McCall wrote:
    Linux isn’t Unix. They appear to have missed off the most commercially successful Unix...macOS.

    Mentioning macOS is counter productive to the people that want to say
    Unix is dead. So why would they point to the now most popular Unix that
    most people don't realize is Unix?



    --
    Grant. . . .
    unix || die

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Scott Dorsey@21:1/5 to ian@eruvia.org on Wed Feb 22 12:55:29 2023
    Ian McCall <ian@eruvia.org> wrote:
    On 11 Feb 2023, Y A wrote
    (in article<420c786f-934f-4bbc-b4d2-91793af96655n@googlegroups.com>):

    As far as the linux lives, unix will never die. I am using pop linux.

    On Tuesday, January 17, 2023 at 8:15:21 PM UTC+2, Ben Collver wrote:
    Unix is dead. Long live Unix!

    Don't expect to see any more big AIX news. This means the last Unix
    left is... Linux

    Liam Proven
    Tue 17 Jan 2023

    Linux isn’t Unix. They appear to have missed off the most commercially >successful Unix...macOS.

    MacOS X is really nothing like Unix under the hood. It has a UI that sort of looks like Unix until you start looking hard and realizing that the password file is really just the product of a secret hidden database, etc.

    In some ways it's a win, because you get a kernel designed for multimedia stuff combined with a mostly-Unix UI.
    --scott
    --
    "C'est un Nagra. C'est suisse, et tres, tres precis."

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Ian McCall@21:1/5 to Scott Dorsey on Thu Feb 23 09:23:32 2023
    On 22 Feb 2023, Scott Dorsey wrote
    (in article <tt53c1$s8g$1@panix2.panix.com>):

    Ian McCall <ian@eruvia.org> wrote:
    On 11 Feb 2023, Y A wrote
    (in article<420c786f-934f-4bbc-b4d2-91793af96655n@googlegroups.com>):

    As far as the linux lives, unix will never die. I am using pop linux.

    On Tuesday, January 17, 2023 at 8:15:21 PM UTC+2, Ben Collver wrote:
    Unix is dead. Long live Unix!

    Don't expect to see any more big AIX news. This means the last Unix left is... Linux

    Liam Proven
    Tue 17 Jan 2023

    Linux isn’t Unix. They appear to have missed off the most commercially successful Unix...macOS.

    MacOS X is really nothing like Unix under the hood. It has a UI that sort of looks like Unix until you start looking hard and realizing that the password file is really just the product of a secret hidden database, etc.

    In some ways it's a win, because you get a kernel designed for multimedia stuff
    combined with a mostly-Unix UI.
    --Scott

    It’s literally Unix.
    <https://www.opengroup.org/openbrand/register/>

    People confuse what Unix actually is with an arbitrary layout. And I use
    macOS from the command line constantly - have done for years. I do wish they’d make their mind up about /usr/local/bin staying put in upgrades, but otherwise it’s been fine.

    Cheers,
    Ian

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Eli the Bearded@21:1/5 to ian@eruvia.org on Thu Feb 23 20:53:26 2023
    In comp.misc, Ian McCall <ian@eruvia.org> wrote:
    On 22 Feb 2023, Scott Dorsey wrote
    MacOS X is really nothing like Unix under the hood. It has a UI that sort of >> looks like Unix until you start looking hard and realizing that the password >> file is really just the product of a secret hidden database, etc.

    In some ways it's a win, because you get a kernel designed for multimedia
    stuff combined with a mostly-Unix UI.
    It’s literally Unix.

    I'll leave the full debate on thar for others, but Scott and I have both
    used literal Apple Unix in the form of A/UX and this OSX thing is a
    different beast.

    People confuse what Unix actually is with an arbitrary layout. And I
    use macOS from the command line constantly - have done for years. I do
    wish they’d make their mind up about /usr/local/bin staying put in upgrades, but otherwise it’s been fine.

    As a mostly Linux user these days, with some Mac for $WORK and providing
    tech suport to my family, I find it slightly infuriating to use either
    the gui or the command line, for different reasons.

    The case sensitive filesystem is one of the more annoying aspects from
    the command line. A system wide preference for Capital letters for
    Folders (directories) and for putting spaces in file names makes
    everything a bit trickier. (Linux lets me rename "standard" XDG crap directories with a ~/.config/user-dirs.dirs file, but Apple won't let me
    `mv Desktop desktop` : "Permission denied".)

    Elijah
    ------
    suspects Scott's criteria for Unix are different than typical Linux users

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Rich@21:1/5 to Eli the Bearded on Thu Feb 23 21:43:37 2023
    Eli the Bearded <*@eli.users.panix.com> wrote:
    The case sensitive filesystem is one of the more annoying aspects from
    the command line.

    With Bash on Linux you can turn off a bunch of the case sensitivity in
    the shell if you want. See the "completion-ignore-case" readline
    setting and the "nocaseglob" shopt option.

    This does not change the filesystem's case sensitivity, but it does
    mean you don't have to type the correct case to get a match when using
    the CLI.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Eli the Bearded@21:1/5 to rich@example.invalid on Fri Feb 24 00:06:02 2023
    In comp.misc, Rich <rich@example.invalid> wrote:
    With Bash on Linux you can turn off a bunch of the case sensitivity in
    the shell if you want. See the "completion-ignore-case" readline
    setting and the "nocaseglob" shopt option.

    That's a funny way to say I can turn off case sensitivity in Bash on
    Macs. And while that works, mostly, if and when I'm using bash (/bin/ksh
    is my shell), it's a bandaid becauase that's just the interactive shell,
    not all tools. Commands cut-n-pasted from the command line might not
    work in shell scripts or Makefiles.

    (I can't use the built-in filename precedence of make to have 'makefile' override 'Makefile' either.)

    And the Mac /bin/bash warns you not to use it: "The default interactive
    shell is now zsh." And then instructions on using chsh and link to a
    webpage saying the same thing in more words.

    Elijah
    ------
    prefers the vi mode of ksh over bash

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Ian McCall@21:1/5 to Eli the Bearded on Sat Feb 25 16:08:03 2023
    On 23 Feb 2023, Eli the Bearded wrote
    (in article <eli$2302231519@qaz.wtf>):

    In comp.misc, Ian McCall<ian@eruvia.org> wrote:
    It’s literally Unix.

    I'll leave the full debate on thar for others, but Scott and I have both
    used literal Apple Unix in the form of A/UX and this OSX thing is a
    different beast.

    As have I, and the half-way house “let’s pretend we’re running Unix"
    MPW as well. There’s no getting round this. Unix is defined by the Open Group. The Open Group certify macOS as Unix. macOS is Unix.

    The case sensitive filesystem is one of the more annoying aspects from
    the command line. A system wide preference for Capital letters for
    Folders (directories) and for putting spaces in file names makes
    everything a bit trickier. (Linux lets me rename "standard" XDG crap directories with a ~/.config/user-dirs.dirs file, but Apple won't let me
    `mv Desktop desktop` : "Permission denied".)

    And this is exactly what I mean. None of that is required for a thing to be Unix, it’s an attribute of the filesystem etc..

    Recall this thread started with a reference to an article “Unix is dead” and talking about AIX. AIX is different again and isn’t XDG-compliant to my knowledge at least (certainly never used to be, but I must admit it’s quite
    a while since I used AIX.

    Elijah
    ------
    suspects Scott's criteria for Unix are different than typical Linux users

    Linux isn’t Unix. macOS -is-. This is a pure trademark matter and not
    POSIX, filesystems, Gnu shell capabilities, XDG standards...

    The article was just flat out wrong. Unix is very much not dead. It sells by the tons, in the form of Apple’s macOS.

    Cheers,
    Ian

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Eli the Bearded@21:1/5 to ian@eruvia.org on Sun Feb 26 03:51:46 2023
    In comp.misc, Ian McCall <ian@eruvia.org> wrote:
    Unix is defined by the Open Group. The Open Group certify macOS as
    Unix. macOS is Unix.

    Yes, but what is that definition versus what do people think of as Unix?
    A system with a minimum set of programs (working a certain way),
    directory layout, header files with the right things defined. That's
    Open Group certifiable, but it's not necessary very Unix-y. There are
    loopholes for allowing differing results based on how filesystems work.
    And that's what happens when Apple ships a case insensitive filesystem
    in keeping with how Macs have always worked.

    But the more you poke at Mac OS the more you find ways it has divirged
    from what other Unixes do outside the strict legal limits of what Open
    Group says is Unix. All Unixes are outliers in some aspects. Ones that
    are outliers in many get snubbed as bad Unix (even if they are great at
    being what they set out to be).

    And this is exactly what I mean. None of that is required for a thing
    to be Unix, it's an attribute of the filesystem etc..

    And I fully understand "it's not required" but it definitaly impacts my
    finding it easy/normal to use via command line.

    Elijah
    ------
    doesn't feel Apple really wants to more than "checklist Unix"

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Ben Collver@21:1/5 to Eli the Bearded on Sun Feb 26 14:37:27 2023
    On 2023-02-26, Eli the Bearded <*@eli.users.panix.com> wrote:
    doesn't feel Apple really wants to more than "checklist Unix"

    I like that phrase "checklist Unix."

    When i think of Unix, i think of the New Jersey license plate:

    "UNIX: live free or die"

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Mike Spencer@21:1/5 to Ben Collver on Sun Feb 26 14:09:56 2023
    Ben Collver <bencollver@tilde.pink> writes:

    On 2023-02-26, Eli the Bearded <*@eli.users.panix.com> wrote:

    doesn't feel Apple really wants to more than "checklist Unix"

    I like that phrase "checklist Unix."

    When i think of Unix, i think of the New Jersey license plate:

    "UNIX: live free or die"

    You're thinking of New Hampshire where people truly dedicated to the
    slogan go to jail for refusing to allow the gummint to force them to
    exhibit a political slogan, even one they believe in, on their cars.

    I thought New Jersey's was "Where's the vig?"

    --
    Mike Spencer Nova Scotia, Canada

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Ian McCall@21:1/5 to Mike Spencer on Sun Feb 26 19:07:37 2023
    On 26 Feb 2023, Mike Spencer wrote
    (in article <87y1okfgcr.fsf@bogus.nodomain.nowhere>):


    Ben Collver <bencollver@tilde.pink> writes:

    On 2023-02-26, Eli the Bearded<*@eli.users.panix.com> wrote:

    doesn't feel Apple really wants to more than "checklist Unix"

    I like that phrase "checklist Unix."

    When i think of Unix, i think of the New Jersey license plate:

    "UNIX: live free or die"

    You're thinking of New Hampshire where people truly dedicated to the
    slogan go to jail for refusing to allow the gummint to force them to
    exhibit a political slogan, even one they believe in, on their cars.

    I thought New Jersey's was "Where's the vig?”

    It was a little bit both when it comes to the certification. Apple wasn’t certified to be Unix but kept saying it was whereas Open Group realised more people cared about Apple than them.

    Interesting info from the horse’s mouth, so to speak: <https://www.quora.com/What-goes-into-making-an-OS-to-be-Unix-compliant- certified>

    Cheers,
    Ian

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Ian McCall@21:1/5 to All on Sun Feb 26 19:12:34 2023
    On 26 Feb 2023, Ian McCall wrote
    (in article<0001HW.29ABE57901A6AD5870000FCBD38F@news.individual.net>):

    Interesting info from the horse’s mouth, so to speak: <https://www.quora.com/What-goes-into-making-an-OS-to-be-Unix-compliant- certified>

    Relevant to, to anyone that’s coded in the Linux kernel before:

    "If I were asked to do the same thing for Linux, it likely would take five years, and two dozen people. Linux is pretty balkanize, has a lot of kingdom building, and you have to pee on everything to make it smell like Linux.
    I could do the same in FreeBSD in about a year and a half, with a dozen co-conspirators to run the changes through.
    A lot of the work would happen in the “ports” tree.”

    I’ve had the same feedback every time about Linux, and I’ve coded in the kernel myself in the early days as well (~1995-1997’ish). Very similar experience with Mozilla too when I tried to add a feature to Thunderbird. Got so put off I just dumped the idea of helping out completely.

    I spoke with the Sun people who designed ZFS and also D. They said they explicitly designed it to be compatible with Linux, including the licensing, and they just got a lot of Not Invented Here chat back. That’s been my experience when dealing with the less pretty guts of Linux too.

    Cheers,
    Ian

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Ben Collver@21:1/5 to Mike Spencer on Mon Feb 27 05:26:59 2023
    On 2023-02-26, Mike Spencer <mds@bogus.nodomain.nowhere> wrote:

    Ben Collver <bencollver@tilde.pink> writes:

    On 2023-02-26, Eli the Bearded <*@eli.users.panix.com> wrote:

    doesn't feel Apple really wants to more than "checklist Unix"

    When i think of Unix, i think of the New Jersey license plate:

    "UNIX: live free or die"

    You're thinking of New Hampshire...

    You're right, i meant New Hampshire :-}

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Geoff Clare@21:1/5 to Ian McCall on Mon Feb 27 13:30:30 2023
    Ian McCall wrote:

    Unix is defined by the Open Group. The Open Group certify macOS
    as Unix. macOS is Unix.

    [...]

    Linux isn’t Unix. macOS -is-. This is a pure trademark matter and not POSIX, filesystems, Gnu shell capabilities, XDG standards...

    You are right about the other things, but not POSIX. To achieve
    UNIX certification (and thus be able to use the trademark) requires
    passing tens of thousands of tests which check that the system behaves
    as per the requirements of POSIX.1 and The Single UNIX Specification
    (which are one and the same document).

    --
    Geoff Clare <netnews@gclare.org.uk>

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Geoff Clare@21:1/5 to Ian McCall on Mon Feb 27 14:04:03 2023
    Ian McCall wrote:

    Apple wasn’t certified to be Unix but kept saying it was whereas
    Open Group realised more people cared about Apple than them.

    Interesting info from the horse’s mouth, so to speak: <https://www.quora.com/What-goes-into-making-an-OS-to-be-Unix-compliant- certified>

    That article paints a completely different picture than your
    characterisation of what happened as "Open Group realised more people
    cared about Apple than them".

    It basically says that Apple had two choices to get out of the
    lawsuit: 1. Extensively modify macOS so that it could be certified,
    or 2. Buy The Open Group (and thus own the trademark). It's pretty
    clear that The Open Group did not make any concessions to Apple at
    all; they just certified macOS the same way they did AIX, HP-UX,
    Solaris, etc. (which, as I said in my other reply, involves passing
    tens of thousands of tests).

    --
    Geoff Clare <netnews@gclare.org.uk>

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Ian McCall@21:1/5 to Geoff Clare on Mon Feb 27 23:24:18 2023
    On 27 Feb 2023, Geoff Clare wrote
    (in article<msnucj-3hi.ln1@ID-313840.user.individual.net>):

    Ian McCall wrote:

    Unix is defined by the Open Group. The Open Group certify macOS
    as Unix. macOS is Unix.
    [...]

    Linux isn’t Unix. macOS -is-. This is a pure trademark matter and not POSIX, filesystems, Gnu shell capabilities, XDG standards...

    You are right about the other things, but not POSIX. To achieve
    UNIX certification (and thus be able to use the trademark) requires
    passing tens of thousands of tests which check that the system behaves
    as per the requirements of POSIX.1 and The Single UNIX Specification
    (which are one and the same document).

    Windows NT was POSIX compliant. MVS was POSIX. And neither were Unix.
    They’re different specs.

    Cheers,
    Ian

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Scott Dorsey@21:1/5 to ian@eruvia.org on Tue Feb 28 02:54:11 2023
    Ian McCall <ian@eruvia.org> wrote:

    Relevant to, to anyone that’s coded in the Linux kernel before:

    "If I were asked to do the same thing for Linux, it likely would take five >years, and two dozen people. Linux is pretty balkanize, has a lot of kingdom >building, and you have to pee on everything to make it smell like Linux.
    I could do the same in FreeBSD in about a year and a half, with a dozen >co-conspirators to run the changes through.

    This is a good thing. I WANT it to be hard to add features to the kernel.
    I am tired of people adding features to the kernel. Please stop adding features.

    I would in fact claim that the BSD kernel is comparatively less feature-ridden in spite of their policies.
    --scott

    --
    "C'est un Nagra. C'est suisse, et tres, tres precis."

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Scott Dorsey@21:1/5 to *@eli.users.panix.com on Tue Feb 28 02:51:23 2023
    Eli the Bearded <*@eli.users.panix.com> wrote:
    In comp.misc, Ian McCall <ian@eruvia.org> wrote:
    Unix is defined by the Open Group. The Open Group certify macOS as
    Unix. macOS is Unix.

    Yes, but what is that definition versus what do people think of as Unix?

    If it does everything v7 does the way v7 does it, it's effectively Unix. --scott

    --
    "C'est un Nagra. C'est suisse, et tres, tres precis."

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Dan Espen@21:1/5 to Ian McCall on Mon Feb 27 22:41:57 2023
    Ian McCall <ian@eruvia.org> writes:

    On 27 Feb 2023, Geoff Clare wrote
    (in article<msnucj-3hi.ln1@ID-313840.user.individual.net>):

    Ian McCall wrote:

    Unix is defined by the Open Group. The Open Group certify macOS
    as Unix. macOS is Unix.
    [...]

    Linux isn’t Unix. macOS -is-. This is a pure trademark matter and not
    POSIX, filesystems, Gnu shell capabilities, XDG standards...

    You are right about the other things, but not POSIX. To achieve
    UNIX certification (and thus be able to use the trademark) requires
    passing tens of thousands of tests which check that the system behaves
    as per the requirements of POSIX.1 and The Single UNIX Specification
    (which are one and the same document).

    Windows NT was POSIX compliant. MVS was POSIX. And neither were Unix. They’re different specs.

    MVS has a unix subsystem. I found it a reasonable unix.

    --
    Dan Espen

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Ian McCall@21:1/5 to Scott Dorsey on Tue Feb 28 07:37:32 2023
    On 28 Feb 2023, Scott Dorsey wrote
    (in article <ttjqcj$g9c$1@panix2.panix.com>):

    Ian McCall <ian@eruvia.org> wrote:

    Relevant to, to anyone that’s coded in the Linux kernel before:

    "If I were asked to do the same thing for Linux, it likely would take five years, and two dozen people. Linux is pretty balkanize, has a lot of kingdom
    building, and you have to pee on everything to make it smell like Linux.
    I could do the same in FreeBSD in about a year and a half, with a dozen co-conspirators to run the changes through.

    This is a good thing. I WANT it to be hard to add features to the kernel.
    I am tired of people adding features to the kernel. Please stop adding features.

    I would in fact claim that the BSD kernel is comparatively less feature-ridden
    in spite of their policies.
    --Scott
    Couldn’t agree more.

    Do they have a stable ABI yet? When I was coding it, they didn’t and touted that as an -advantage-. It wasn’t of course, dreadful idea and the reason drivers kept needed recompiling all the time. Reminded me of the expunged-from-history page the MySQL site used to have explaining that
    foreign keys were bad “because it makes application programming hard”.

    <sigh>.

    Cheers,
    Ian

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Ian McCall@21:1/5 to Geoff Clare on Tue Feb 28 13:43:12 2023
    On 28 Feb 2023, Geoff Clare wrote
    (in article<9dc1dj-mdg.ln1@ID-313840.user.individual.net>):

    Ian McCall wrote:

    On 27 Feb 2023, Geoff Clare wrote
    (in article<msnucj-3hi.ln1@ID-313840.user.individual.net>):

    Ian McCall wrote:

    Linux isn’t Unix. macOS -is-. This is a pure trademark matter and not POSIX, filesystems, Gnu shell capabilities, XDG standards...

    You are right about the other things, but not POSIX. To achieve
    UNIX certification (and thus be able to use the trademark) requires passing tens of thousands of tests which check that the system behaves
    as per the requirements of POSIX.1 and The Single UNIX Specification (which are one and the same document).

    Windows NT was POSIX compliant. MVS was POSIX. And neither were Unix. They’re different specs.

    Windows NT and MVS were certified POSIX compliant in the mid 1990's.
    At that time POSIX.1 and SUS were indeed different specs, and POSIX.2
    was separate from POSIX.1. All three were merged in 2001 to form POSIX.1-2001/SUSv3, which was before macOS was certified.

    Interesting - thanks for that.

    Cheers,
    Ian

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Geoff Clare@21:1/5 to Ian McCall on Tue Feb 28 13:32:57 2023
    Ian McCall wrote:

    On 27 Feb 2023, Geoff Clare wrote
    (in article<msnucj-3hi.ln1@ID-313840.user.individual.net>):

    Ian McCall wrote:

    Linux isn’t Unix. macOS -is-. This is a pure trademark matter and not
    POSIX, filesystems, Gnu shell capabilities, XDG standards...

    You are right about the other things, but not POSIX. To achieve
    UNIX certification (and thus be able to use the trademark) requires
    passing tens of thousands of tests which check that the system behaves
    as per the requirements of POSIX.1 and The Single UNIX Specification
    (which are one and the same document).

    Windows NT was POSIX compliant. MVS was POSIX. And neither were Unix. They’re different specs.

    Windows NT and MVS were certified POSIX compliant in the mid 1990's.
    At that time POSIX.1 and SUS were indeed different specs, and POSIX.2
    was separate from POSIX.1. All three were merged in 2001 to form POSIX.1-2001/SUSv3, which was before macOS was certified.

    --
    Geoff Clare <netnews@gclare.org.uk>

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Dan Cross@21:1/5 to ian@eruvia.org on Tue Feb 28 16:59:28 2023
    In article <0001HW.29ADE6BC01D1602C70000FCBD38F@news.individual.net>,
    Ian McCall <ian@eruvia.org> wrote:
    On 28 Feb 2023, Scott Dorsey wrote
    (in article <ttjqcj$g9c$1@panix2.panix.com>):

    Ian McCall <ian@eruvia.org> wrote:

    Relevant to, to anyone that’s coded in the Linux kernel before:

    "If I were asked to do the same thing for Linux, it likely would take five >> > years, and two dozen people. Linux is pretty balkanize, has a lot of kingdom
    building, and you have to pee on everything to make it smell like Linux. >> > I could do the same in FreeBSD in about a year and a half, with a dozen
    co-conspirators to run the changes through.

    This is a good thing. I WANT it to be hard to add features to the kernel.
    I am tired of people adding features to the kernel. Please stop adding
    features.

    I would in fact claim that the BSD kernel is comparatively less feature-ridden
    in spite of their policies.
    --Scott
    Couldn't agree more.

    Do they have a stable ABI yet? When I was coding it, they didn't and touted >that as an -advantage-. It wasn't of course, dreadful idea and the reason >drivers kept needed recompiling all the time. Reminded me of the >expunged-from-history page the MySQL site used to have explaining that >foreign keys were bad "because it makes application programming hard".

    I think you mean stable internal interfaces, not ABI.
    One of the Linux invariants is, "never break userspace!"
    which implies a stable system call interface.

    Of course, they change the format of files under /sys
    and stuff, but hey.

    - Dan C.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Bob Eager@21:1/5 to Ian McCall on Wed Mar 1 09:07:20 2023
    On Wed, 01 Mar 2023 09:02:40 +0000, Ian McCall wrote:

    On 28 Feb 2023, Dan Cross wrote (in article <ttlbtg$iql$1@reader2.panix.com>):

    In article<0001HW.29ADE6BC01D1602C70000FCBD38F@news.individual.net>,
    Ian McCall <ian@eruvia.org> wrote:
    Do they have a stable ABI yet? When I was coding it, they didn't and
    touted that as an -advantage-. It wasn't of course, dreadful idea and
    the reason drivers kept needed recompiling all the time. Reminded me
    of the expunged-from-history page the MySQL site used to have
    explaining that foreign keys were bad "because it makes application
    programming hard".

    I think you mean stable internal interfaces, not ABI.
    One of the Linux invariants is, "never break userspace!"
    which implies a stable system call interface.

    Of course, they change the format of files under /sys and stuff, but
    hey.

    Definitely meant ABI. It’s the binary bit that’s key - it’s why people had to recompile drivers all the time which didn’t happen with other
    OSs.

    For quite a while, that only has to be done with a new major version.

    --
    Using UNIX since v6 (1975)...

    Use the BIG mirror service in the UK:
    http://www.mirrorservice.org

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Ian McCall@21:1/5 to Dan Cross on Wed Mar 1 09:02:40 2023
    On 28 Feb 2023, Dan Cross wrote
    (in article <ttlbtg$iql$1@reader2.panix.com>):

    In article<0001HW.29ADE6BC01D1602C70000FCBD38F@news.individual.net>,
    Ian McCall <ian@eruvia.org> wrote:
    Do they have a stable ABI yet? When I was coding it, they didn't and touted that as an -advantage-. It wasn't of course, dreadful idea and the reason drivers kept needed recompiling all the time. Reminded me of the expunged-from-history page the MySQL site used to have explaining that foreign keys were bad "because it makes application programming hard".

    I think you mean stable internal interfaces, not ABI.
    One of the Linux invariants is, "never break userspace!"
    which implies a stable system call interface.

    Of course, they change the format of files under /sys
    and stuff, but hey.

    Definitely meant ABI. It’s the binary bit that’s key - it’s why people had to recompile drivers all the time which didn’t happen with other OSs.

    Cheers,
    Ian

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Ian McCall@21:1/5 to Bob Eager on Wed Mar 1 12:11:08 2023
    On 1 Mar 2023, Bob Eager wrote
    (in article <k68iu8FugeuU10@mid.individual.net>):

    On Wed, 01 Mar 2023 09:02:40 +0000, Ian McCall wrote:

    On 28 Feb 2023, Dan Cross wrote (in article <ttlbtg$iql$1@reader2.panix.com>):

    In article<0001HW.29ADE6BC01D1602C70000FCBD38F@news.individual.net>,
    Ian McCall <ian@eruvia.org> wrote:
    Do they have a stable ABI yet? When I was coding it, they didn't and touted that as an -advantage-. It wasn't of course, dreadful idea and the reason drivers kept needed recompiling all the time. Reminded me
    of the expunged-from-history page the MySQL site used to have explaining that foreign keys were bad "because it makes application programming hard".

    I think you mean stable internal interfaces, not ABI.
    One of the Linux invariants is, "never break userspace!"
    which implies a stable system call interface.

    Of course, they change the format of files under /sys and stuff, but
    hey.

    Definitely meant ABI. It’s the binary bit that’s key - it’s why people
    had to recompile drivers all the time which didn’t happen with other
    OSs.

    For quite a while, that only has to be done with a new major version.

    Thanks. Hard to find exactly when it got one, but this article seems to
    suggest it was around 2016: <https://www.phoronix.com/news/Linux-Kernel-Stable-API-ABI>

    ...so yes, that’s after I was doing kernel coding by quite a way.

    Cheers,
    Ian

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Dan Cross@21:1/5 to ian@eruvia.org on Wed Mar 1 15:01:56 2023
    In article <0001HW.29AF4C30020D4BA470000FCBD38F@news.individual.net>,
    Ian McCall <ian@eruvia.org> wrote:
    On 28 Feb 2023, Dan Cross wrote
    (in article <ttlbtg$iql$1@reader2.panix.com>):

    In article<0001HW.29ADE6BC01D1602C70000FCBD38F@news.individual.net>,
    Ian McCall <ian@eruvia.org> wrote:
    Do they have a stable ABI yet? When I was coding it, they didn't and touted
    that as an -advantage-. It wasn't of course, dreadful idea and the reason >> > drivers kept needed recompiling all the time. Reminded me of the
    expunged-from-history page the MySQL site used to have explaining that
    foreign keys were bad "because it makes application programming hard".

    I think you mean stable internal interfaces, not ABI.
    One of the Linux invariants is, "never break userspace!"
    which implies a stable system call interface.

    Of course, they change the format of files under /sys
    and stuff, but hey.

    Definitely meant ABI. It's the binary bit that's key - it's why people
    had to recompile drivers all the time which didn't happen with other OSs.

    Do you, though? Oh, say, x86, Linux uses the SysV ABI. People
    need to recompile their drivers because the internals change all
    the time, not because the calling convention and structure
    layouts change.

    - Dan C.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Bob Eager@21:1/5 to Ian McCall on Wed Mar 1 15:30:17 2023
    On Wed, 01 Mar 2023 12:11:08 +0000, Ian McCall wrote:

    On 1 Mar 2023, Bob Eager wrote (in article <k68iu8FugeuU10@mid.individual.net>):

    On Wed, 01 Mar 2023 09:02:40 +0000, Ian McCall wrote:

    On 28 Feb 2023, Dan Cross wrote (in article
    <ttlbtg$iql$1@reader2.panix.com>):

    In
    article<0001HW.29ADE6BC01D1602C70000FCBD38F@news.individual.net>,
    Ian McCall <ian@eruvia.org> wrote:
    Do they have a stable ABI yet? When I was coding it, they didn't
    and touted that as an -advantage-. It wasn't of course, dreadful
    idea and the reason drivers kept needed recompiling all the time.
    Reminded me of the expunged-from-history page the MySQL site used
    to have explaining that foreign keys were bad "because it makes
    application programming hard".

    I think you mean stable internal interfaces, not ABI.
    One of the Linux invariants is, "never break userspace!"
    which implies a stable system call interface.

    Of course, they change the format of files under /sys and stuff,
    but hey.

    Definitely meant ABI. It’s the binary bit that’s key - it’s why
    people had to recompile drivers all the time which didn’t happen with
    other OSs.

    For quite a while, that only has to be done with a new major version.

    Thanks. Hard to find exactly when it got one, but this article seems to suggest it was around 2016: <https://www.phoronix.com/news/Linux-Kernel-Stable-API-ABI>

    ...so yes, that’s after I was doing kernel coding by quite a way.

    Sorry, I was referring to FreeBSD!



    --
    Using UNIX since v6 (1975)...

    Use the BIG mirror service in the UK:
    http://www.mirrorservice.org

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Ian McCall@21:1/5 to Dan Cross on Wed Mar 1 16:58:31 2023
    On 1 Mar 2023, Dan Cross wrote
    (in article <ttnpd4$7l7$3@reader2.panix.com>):

    Do you, though? Oh, say, x86, Linux uses the SysV ABI. People
    need to recompile their drivers because the internals change all
    the time, not because the calling convention and structure
    layouts change.

    The only reason you’d need to recompile drivers as opposed to install binaries is because of the ABI. It’s possible you might -choose- to, but
    the only reason you’d need to is ABI. This is trivially provable by the
    fact that every other OS, which has a stable ABI, doesn’t require you to compile your drivers but can instead install them as binaries.

    Linux used to tout this as an advantage (“used to” - remember I was
    mostly coding in the kernel in the 90s, so don’t know when or if they
    changed their opinion). It isn’t an advantage, it’s a -disadvantage-. Hopefully they changed their opinion.

    Cheers,
    Ian

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Scott Dorsey@21:1/5 to dan1espen@gmail.com on Wed Mar 1 19:24:51 2023
    Dan Espen <dan1espen@gmail.com> wrote:

    MVS has a unix subsystem. I found it a reasonable unix.

    I can't even imagine that. Using vi on a 3270 is farther than my mind can stretch.

    I did try NOS VE/VX on the CDC Cyber machines.... very interesting having to type *EOI instead of ctrl-D. And amazing using vi on a system where the host I/O is line oriented and doesn't see anything until you hit return... the
    PPU does all the hard lifting.
    --scott


    --
    "C'est un Nagra. C'est suisse, et tres, tres precis."

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Dan Espen@21:1/5 to Scott Dorsey on Wed Mar 1 15:25:46 2023
    kludge@panix.com (Scott Dorsey) writes:

    Dan Espen <dan1espen@gmail.com> wrote:

    MVS has a unix subsystem. I found it a reasonable unix.

    I can't even imagine that. Using vi on a 3270 is farther than my mind can stretch.

    Yeah, no.

    You can, of course, do some things on a 3270, but you're better off
    using a terminal of some type. You can also get X-windows going on the mainframe and use MVS that way.

    --
    Dan Espen

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Dan Cross@21:1/5 to ian@eruvia.org on Wed Mar 1 22:09:34 2023
    In article <0001HW.29AFBBB702276F2C70000FCBD38F@news.individual.net>,
    Ian McCall <ian@eruvia.org> wrote:
    On 1 Mar 2023, Dan Cross wrote
    (in article <ttnpd4$7l7$3@reader2.panix.com>):

    Do you, though? Oh, say, x86, Linux uses the SysV ABI. People
    need to recompile their drivers because the internals change all
    the time, not because the calling convention and structure
    layouts change.

    The only reason you'd need to recompile drivers as opposed to install >binaries is because of the ABI. It's possible you might -choose- to, but
    the only reason you'd need to is ABI.

    What, exactly, is your definition of "the ABI"?

    This is trivially provable by the
    fact that every other OS, which has a stable ABI, doesn't require you to >compile your drivers but can instead install them as binaries.

    You seem to be conflating an "ABI" with the general notion of a
    programming interface. I agree that Linux does not keep the
    latter stable within the kernel, but that is not the same as the
    former.

    Linux used to tout this as an advantage ("used to" - remember I was
    mostly coding in the kernel in the 90s, so don't know when or if they
    changed their opinion). It isn't an advantage, it's a -disadvantage-. >Hopefully they changed their opinion.

    Perhaps you are referring to this document from gregkh: https://www.kernel.org/doc/html/v5.18/process/stable-api-nonsense.html

    Lack of a stable binary _kernel_ interface for drivers is why
    you have to rebuild drivers on Linux; this has very little to do
    with the ABI they use on any platform, which _is_ well-defined.
    For example, on x86_64 Linux uses the SystemV ABI, and from the
    SYSV ABI document, we get this definition:

    "The System V Application Binary Interface, or ABI,
    defines a system interface for compiled application
    programs and a minimal environment for support of
    installation scripts. Its purpose is to document
    a standard binary interface for application programs
    on systems that implement an operating system that
    complies with the X/Open Common Application
    Environment Specification, Issue 4.2 and the System
    V Interface Definition, Fourth Edition."

    (https://www.sco.com/developers/devspecs/gabi41.pdf)

    Note that this discusses _application_ programs, which device
    drivers on Linux most certainly are not. More generally, ABIs
    are used to specify things like executable file formats (e.g.,
    ELF), function calling conventions (how registers and the stack
    are used, etc), and how data structures are laid out in memory
    (e.g., how the elements of a `C` struct are ordered in memory,
    aligned to various addresses, etc). That's not the issue here,
    nor why you need to recompile Linux drivers when the kernel is
    updated.

    The issue with Linux is that they don't give you stable internal
    kernel interfaces to program against, as described in the greghk
    document linked above. In particular, Linux reserves the right
    to change around the signatures of internal functions and so on
    to suit themselves, deprecate old things, introduce new ones,
    etc; that's what they mean. But that has nothing to do with the
    ABI that compilers and assembler programmers use. Put another
    way, the functions may change, but the way you call them (first
    argument in `%rdi`, second in `%rsi`, etc; stack aligned to a
    16-byte boundary before the `CALL` instruction; the format of
    the `va_list` for variadic functions; etc) does not.

    Compare to, say, the illumos kernel where the DDI gives you
    documented stable interfaces. But again, that's not an ABI;
    that's an internal interface within the kernel. In fact, Linux
    and illumos/Solaris use the same ABI on x86_64.

    - Dan C.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Richard Kettlewell@21:1/5 to Dan Cross on Thu Mar 2 08:06:05 2023
    cross@spitfire.i.gajendra.net (Dan Cross) writes:
    Ian McCall <ian@eruvia.org> wrote:
    The only reason you'd need to recompile drivers as opposed to install
    binaries is because of the ABI. It's possible you might -choose- to,
    but the only reason you'd need to is ABI.

    What, exactly, is your definition of "the ABI"?

    This is trivially provable by the fact that every other OS, which has
    a stable ABI, doesn't require you to compile your drivers but can
    instead install them as binaries.

    You seem to be conflating an "ABI" with the general notion of a
    programming interface. I agree that Linux does not keep the
    latter stable within the kernel, but that is not the same as the
    former.

    You seem to be using ABI to mean little more than calling conventions
    and object file formats, which is a very narrow usage of the term, and certainly not a universal one.

    A long-standing example which goes beyond that would be IBCS, which
    specified details for syscalls, structure definitions, etc for an ABI
    between a user program and a kernel on an x86 Unix. (I think Linux
    suported it at one point, I don’t know if it still does and it’s barely relevant any more anyway.)

    The Linux kernel’s module interface, from the perspective of compiled
    code, is an ABI in that broader sense (as well as being an API from the perspective of source code).

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

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Dan Cross@21:1/5 to invalid@invalid.invalid on Thu Mar 2 14:03:33 2023
    In article <wwvilfj4lxu.fsf@LkoBDZeT.terraraq.uk>,
    Richard Kettlewell <invalid@invalid.invalid> wrote: >cross@spitfire.i.gajendra.net (Dan Cross) writes:
    Ian McCall <ian@eruvia.org> wrote:
    The only reason you'd need to recompile drivers as opposed to install
    binaries is because of the ABI. It's possible you might -choose- to,
    but the only reason you'd need to is ABI.

    What, exactly, is your definition of "the ABI"?

    This is trivially provable by the fact that every other OS, which has
    a stable ABI, doesn't require you to compile your drivers but can
    instead install them as binaries.

    You seem to be conflating an "ABI" with the general notion of a
    programming interface. I agree that Linux does not keep the
    latter stable within the kernel, but that is not the same as the
    former.

    You seem to be using ABI to mean little more than calling conventions
    and object file formats, which is a very narrow usage of the term, and >certainly not a universal one.

    Yeah, that's pretty much what an "ABI" is.

    What you are suggesting amounts to saying that programmers
    cannot change the order of arguments to a function without
    violating an ABI; taken to its logical conclusion, any notion of
    an interface in a compiled language would then form an "ABI",
    which is ... weird. Particularly if the interface in question
    is not part of an application (note that the "A" in "ABI" stands
    for "application").

    I am unaware of any sense in which an ABI has ever been
    generalized to mean the specific details of, say, _function_
    signatures and return values of internal interfaces. Perhaps
    you could share if you know of one?

    A long-standing example which goes beyond that would be IBCS, which
    specified details for syscalls, structure definitions, etc for an ABI
    between a user program and a kernel on an x86 Unix. (I think Linux
    suported it at one point, I don't know if it still does and it's barely >relevant any more anyway.)

    I think you mean, "iBCS" (note the lower-case "i") and the later
    iBCS2. Too bad implementations had non-standard extensions that
    lead to incompatibilities. It's almost like it wasn't a very
    good standard. Broadly, the System V ABI with the
    processor-specific supplements is equivalent.

    This perhaps gets to the core of the issue; say the system call
    interface provided by Linux is considered stable (one of their
    assurances is to not break userspace by changing it --- at least
    not without a very, very good reason). But that's entirely
    separate from an internal programming interface within the
    kernel, which is what device drivers use.

    Think of it this way: C11 standardizes a library interface in
    the form of the C library. The signature of, say, `memcpy` is
    well-defined. So why can't I compile a bit of code that
    calls `memcpy` on Windows and link it against Linux's libc?
    That's the ABI difference.

    The Linux kernel's module interface, from the perspective of compiled
    code, is an ABI in that broader sense (as well as being an API from the >perspective of source code).

    Nope, sorry.

    The Linux kernel module interface is defined to not be stable;
    what you and the OP are asking for are two separate but related
    things: a stable interface and a stable ABI to call it with.
    You get the latter, but not the former.

    - Dan C.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Richard Kettlewell@21:1/5 to Dan Cross on Thu Mar 2 14:49:50 2023
    cross@spitfire.i.gajendra.net (Dan Cross) writes:
    The Linux kernel module interface is defined to not be stable;
    what you and the OP are asking for are two separate but related
    things: a stable interface and a stable ABI to call it with.
    You get the latter, but not the former.

    I’m not asking for anything, I’m just saying what I think an ABI
    means. Based on e.g. Wikipedia and the Linux kernel docs, I don’t think
    I’m alone.

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

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Dan Cross@21:1/5 to invalid@invalid.invalid on Thu Mar 2 14:52:05 2023
    In article <wwvzg8vmcmp.fsf@LkoBDZeT.terraraq.uk>,
    Richard Kettlewell <invalid@invalid.invalid> wrote: >cross@spitfire.i.gajendra.net (Dan Cross) writes:
    The Linux kernel module interface is defined to not be stable;
    what you and the OP are asking for are two separate but related
    things: a stable interface and a stable ABI to call it with.
    You get the latter, but not the former.

    I'm not asking for anything, I'm just saying what I think an ABI
    means. Based on e.g. Wikipedia and the Linux kernel docs, I don't think
    I'm alone.

    That's odd. What you are suggesting seems to be directly
    contradicted by wikipedia, the Linux kernel docs, and the System
    V ABI document and it's supplements.

    Context matters here. In the context of Linux, lack of stable
    driver interfaces just isn't an ABI issue.

    - Dan C.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Julieta Shem@21:1/5 to Grant Taylor on Sun Dec 3 12:46:56 2023
    Grant Taylor <gtaylor@tnetconsulting.net> writes:

    On 1/18/23 2:58 AM, Marco Moock wrote:
    There is UNIX and UNIX-like. Some people confuse them.

    I usually refer to these as Unix and unix (like OS).

    The original UNIX was spelled with upper case letters. We can write
    Unix for the whole family and UNIX for the root of the tree. Same thing
    we do with LISP and Lisp.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)