[Info-vax] Status of the PostgreSQL port?

[Arne Vajhøj]

On 6/29/2015 6:51 PM, John Reagan wrote:
> On Monday, June 29, 2015 at 2:36:28 PM UTC-4, Stephen Hoffman wrote:
>> On 2015-06-29 17:22:04 +0000, BillPedersen said:
>>> And setting the DECC$ logicals in LIB$INITIALIZE is exactly what
>>> we do to control the behavior based upon where and how the ported
>>> routines are invoked - if necessary.
>>
>> Which is not the most maintainable user interface for programming
>> and for run-time, particularly as feature knobs are added.
>
> OpenVMS is in a position to make itself look like Unix without making
> itself look like Unix.  Softlinks, mount points, etc. are sometimes
> difficult to squeeze into the traditional OpenVMS device/logical-name
> paradigm.  Different programs want different things.
>
> Open Source code doesn't come with OpenVMS-isms, but often finds
> itself at odds with filename syntax differences.
>
> The CRTL has tried to be upward compatible at the expense of
> complexity.  Can I get rid of stuff for x86?  Anybody still using the
> smaller stat for example?

For x86-64 you can obviously get rid of requirements for
binary compatibility.

When it comes to source code and library compatibility then
it becomes more muddy.

I think that a CC/MODE=VMSISH and LINK/MODE=VMSISH and
CC/MODE=PORTABLE and LINK/MODE=PORTABLE would be good.

Then both those porting very old VMS applications and
those porting  various *nix open source would be happy.

Arne