[oi-dev] OI project reboot required

[Piotr Jasiukajtis]

On May 12, 2013, at 7:10 PM, Garrett D'Amore <garrett.damore at dey-sys.com> wrote:

> 
> On May 12, 2013, at 9:02 AM, Jim Klimov <jimklimov at cos.ru> wrote:
> 
>> On 2013-05-12 17:51, Alan Coopersmith wrote:
>>> On 05/12/13 05:19 AM, David Höppner wrote:
>>>> I noticed Oracle upstream moves aggressively to amd64 only;
>>>> installing amd64 just in bin not in bin/$(MACH64).
>>> 
>>> It has been a few years since Oracle upstream dropped 32-bit i386 support,
>>> so that's just one of the decisions OI has to make - track upstream as is
>>> or fork/patch as needed to continue to support 32-bit on i386.
>> 
>> I believe, 32-bit should be retained. While it is of little utility
>> for ZFS and other huge-RAM jobs, it may be required for some netbooks,
>> older hardware repurposed for tests and SOHO servers, as well as for
>> resource-constrained testing VMs. So I'd vouch for this fork/patch
>> approach if this upstream is still followed.
> 
> We've been doing this for years now.  I'm now starting to think -- 3 years later on -- that this argument feels specious today.  Who runs illumos on a netbook?  I did, once.  Not any more.  (And modern netbooks have 64 bit support!)
> 
> Older hardware must be *really* old.  Over 5 years.  For servers, probably over 10 years.  I've thrown away my Pentiums and Pentium IIs.  I suppose there could be some Pentium IIIs and IVs out there, or AMD Athlons (pre-Athlon64), but they'd all be really really slow by today's standards.  Do people run illumos on such kit?  I'm highly doubtful, unless that kit is around just to answer the question of whether 32-bit kernels still work. :-)
I have used 32bit before illumos was born. It was really painful though. 
I think if someone is capable of running illumos based system on 32bit these days, he should also be capable of maintaing 32bit fork for his own needs.

--
Piotr Jasiukajtis