AW: drop ia64 from man pages?

[Walter Harms <wharms@bfs.de>]

>What makes a valid variant though?
>There is no upstream supported glibc older than 2.32.
>There is no upstream supported Linux older than 5.4 (LTS).

Maybe, i am using a LOT of embedded Systems and they are using sometimes
really old stuff. So sometimes i am thankfull for information about older variants.

reminder: man pages are not for server stuff only.
Same goes for older programms, you can only understand when you have the old documentation.

So as long as IA64 is in use, there is a chance some need that info.

________________________________________
Von: Carlos O'Donell <carlos@redhat.com>
Gesendet: Donnerstag, 17. Juli 2025 14:13:54
An: Eugene Syromyatnikov
Cc: enh; Alejandro Colomar; linux-man
Betreff: Re: drop ia64 from man pages?

On 7/17/25 6:30 AM, Eugene Syromyatnikov wrote:
> On Wed, Jul 16, 2025 at 7:43 PM Carlos O'Donell <carlos@redhat.com> wrote:
>>
>> On 7/16/25 12:30 PM, enh wrote:
>>> i didn't look at the other pages, but quite a lot on the clone(2) page
>>> is actually about what glibc does ... but glibc already removed all
>>> this stuff. so it should probably not be more than what we have for,
>>> say, m68k which is just "read your kernel/libc source for more"?
>>>
>>> a corollary to "museum hardware should run museum software" might be
>>> "...and use museum documentation" :-)
>>
>> Agreed.
>>
>> There is a balance here between documentation that covers a reasonable
>> number of use cases, documentation that is easy to maintain, and
>> documentation that is easy to read (without superfluous information,
>> either too new or too old).
>>
>> It has been about 1.5 years since IA64 started being dropped, and I
>> don't see any reason to keep very specific documentation about it
>> around except as smaller interesting historical notes.
>
> Depends on whether man pages limit themselves to reflecting only the
> "current" version (whatever this is, as man-pages is not part of
> either linux or glibc source tree), or strive to provide actual useful
> reference for users of systems that may have different variants and
> versions of the kernel and libc.  If it is the latter, outright
> removal (instead of keeping all the pertinent information in the
> history section) is pretty short-sighted.

(1) Co-evolution.

The Linux man-pages project, and most projects, co-evolve with the
ecosystem.

At any point in time you can take the most pertinent release of a
project and use that. VCS history is available to everyone.

This is how downstream distributions have been evolving and serving
users.

(2) A loose matrix of "supported" (not "current")

The project, as I see it, has been providing useful information for a
loose matrix of supported kernels, supported C libraries (glibc, musl,
bionic), and supported international standards e.g. ISO C, POSIX etc.
along with other APIs from BSD etc.

(3) What is a valid variant?

Once something is deprecated my opinion is that we have a duty to
our users to attempt to cleanup the material and make it easier to
consume with less relevant information moved away from main sections
or pages.

At this point in time I'd say IA64 is deprecated in the current
releases of glibc and linux and so moving the related information,
or cleaning it up seems appropriate. How much of that to do I leave
up to Alejandro as editor (or contributors to work out).

What makes a valid variant though?

There is no upstream supported glibc older than 2.32.

There is no upstream supported Linux older than 5.4 (LTS).

So between the two, there is still IA64 support present.

Supported (all upstream projects support it)
  -> Deprecated (current project releases have removed support)
     -> End of life (no actively maintained project support it)

I think we should cleanup and move content at the "deprecated"
phase, which is where we find IA64 today, and when we get to
EOL, we should be removing all the content related to it except
for historical references that serve an educational or
elucidating purpose.

--
Cheers,
Carlos.