Windows-x86 (32-bit) build is broken. Time to retire it?

[Stewart Addison]

>  I’d imagine there could still be a lot of older systems (and maybe some
embedded case) with 32bit windows still in use which the removal risks some
possible bad vibes for that user base.
> Although I guess if they’re not updating to newer Windows (and hardware),
they may not be upgrading Java without 32bit Windows support either.

The other argument I've heard in the past is that if someone is using a
proprietary application that contains 32-bit JNI libraries then they
wouldn't be able to migrate over to 64-bit. In that situation users may
well be using a newer 64-bit windows/hardware, but still legitimately
running the 32-bit JVM.

But, of course, anyone with such a legacy application that is not yet
available as 64-bit is unlikely to have a strong requirement to push up to
a newer major version of OpenJDK any time soon ... However much we as a
community might like them to. It would certainly be good to know for
certain what people's use cases are for that.

I can't comment specifically on how many 32-bit windows OSs are out there
but looking at the latest download numbers for that last JDK17 release of
Eclipse Temurin, it's roughly ten times fewer 32-bit downloads than 64-bit
ones, but that means a surprising (at least to me) number of people seem to
still be actively seeking out and selecting that one in preference for some
reason. Although I don't recall seeing anyone pushing back on us (Temurin)
for builds on that platform for JDK21 yet.

> The last supported 32-bit Windows goes out of support in October 2025, if
I'm not mistaken.

Yeah that's my understanding too for Windows 10. On the server OS side 2008
was the last version available in 32-bit I believe and that's already out
of support.

Regards,

Stewart...
--


On Thu, 7 Dec 2023 at 20:37, Eric Bresie <ebresie at gmail.com> wrote:

> Just curious if anyone has any visibility as to how widely 32-bit windows
> may still be in use?
> I’d imagine there could still be a lot of older systems (and maybe some
> embedded case) with 32bit windows still in use which the removal risks some
> possible bad vibes for that user base.
>
> Although I guess if they’re not updating to newer Windows (and hardware),
> they may not be upgrading Java without 32bit Windows support either.
>
> On Thu, Dec 7, 2023 at 5:38 AM Magnus Ihse Bursie <
> magnus.ihse.bursie at oracle.com> wrote:
>
>> On 2023-12-06 11:22, Stewart Addison wrote:
>>
>> >
>> > I think there's a discussion to be had here regarding 32-bit support
>> > in general going forward - I have a specific question about arm32 and
>> > whether anyone is actively maintaining that one but perhaps that's a
>> > matter for another thread unless anyone wants to jump in and say
>> > they'll be keeping arm32 alive in the codebase.
>>
>> Obviously the glory days of 32-bit support are over. But I'm guessing it
>> will be some more time before all 32-bit code can be dropped. So I think
>> it is important not to conflate dropping general 32-bit support with the
>> more specific case of dropping win32. The former is more about cpu
>> support, the latter more about os support.
>>
>> Arm32 in specific has never had any really active backers; it seems to
>> be slowly dying but is apparently still on life-support. Let's keep that
>> out of the win32 discussion.
>>
>> /Magnus
>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.org/pipermail/jdk-dev/attachments/20231208/6f5aa8db/attachment.htm>