Hello,

------- Original Message -------
On Monday, July 17th, 2023 at 7:49 PM, Denis 'GNUtoo' Carikli 
<gnu...@cyberdimension.org> wrote:


> And for the record here's how I use a different architecture: I define
> a package:
> 
> > (define postgresql-14-i686-linux
> > (package
> > (inherit postgresql-14)
> > (name "postgresql-14-i686-linux")
> > (arguments
> > (ensure-keyword-arguments
> > (package-arguments postgresql-14)
> > '(#:system "i686-linux")))))
> 
> 
> And then use it in the PostgreSQL service:
> 
> > (service
> > postgresql-service-type
> > (postgresql-configuration
> > (postgresql
> > (if (target-x86-64?)
> > postgresql-14-i686-linux
> > postgresql-14))
> > [...]))
> 
> 
> The downside is that it prints a warning during boot if I recall
> well but it works fine.
> 
> I'm unsure if that helps the conversation or not though as you might
> want something cleaner.

This is quite similar to what I have at the moment; I copied 
gnu/services/databases.scm locally and removed the account creation from it.  
That works fine, but instead of maintaining my own copy of that file, I'd 
rather submit a patch to have this done nicely in the upstream version.

If I use the unmodified service and with the (operating-system (user ...)) 
entry, it works sometimes, but not reliably due to having two conflicting 
entries for the 'postgres' user.

Best regards,

Martin

Reply via email to