Re: Areas for Solaris support modernization

On Thu, Mar 12, 2026, at 5:19 PM, Greg Burd wrote:
> On Thu, Mar 12, 2026, at 11:07 AM, Tom Lane wrote:
>> "Greg Burd" <greg@burd.me> writes:
>>> errfinish+0x288(1008310c9, 70f, 100832730, 100bff148, 100bff208, 0)
>>> ...
>>> aclitemout+0xf4(100cfa668, 1, 0, 100cfa10f, 100cfa600, 100cf9960)
>>> FunctionCall1Coll+0x2c(100cfa120, 0, 100cfa600, 0, 100cfa10f, 7)
>>> array_out+0x700(2, 0, 100cfa610, 100cfa650, 100cfa638, 0)
>>> FunctionCall1Coll+0x2c(ffffffff7fffed00, 0, 100cfa5e8, 0, 100bff150, 2ef)
>>> OidOutputFunctionCall+0x18(2ef, 100cfa5e8, 409, 100c01150, 100, ffffffff7fffed00)
>>> InsertOneValue+0x2a4(100cfa058, e8, 1009a6f60, 100ccdb84, 40a, 100be5388)
>>
>> Oh ... you are running it at max debug level, aren't you.
>> So InsertOneValue tries to print out the inserted value,
>> and that doesn't work because I didn't hot-wire aclitemout
>> for bootstrap mode, only aclitemin.
>
> Hey Tom,
>
> I'm not sure what "hot-wire" means, but I put a quick patch together 
> that sets it to NULL when bootstrapping for now.  I'm building and 
> testing now with the *three* patches, hopefully it'll complete in < 
> 4hrs.

That built fine and test-world passed.  I'll see if I can get the build-farm scripts to run against my local
branch/changesto see if that too passes.  Fingers crossed.
 

-greg

>> I'm away on vacation, but will fix that when I return.
>> In the meantime, backing down initdb's debug level to
>> something less than DEBUG4 should get you past that.
>
> Enjoy the vacation, please ignore me and this issue until you get back. 
>  The tests might be done by then...
>
> best.
>
> -greg
>
>>             regards, tom lane
> Attachments:
> * v20260312c-0001-Restore-PS_USE_CHANGE_ARGV-for-Solaris.patch
> * v20260312c-0002-Switch-the-semaphore-API-on-Solaris-to-un.patch
> * v20260312c-0003-Make-aclitemout-work-during-bootstrap-mod.patch