Re: [PATCH] Proof of concept for GUC improvements

On 8/20/21 12:09 AM, David Christensen wrote:
> -hackers,
> 
> Enclosed, find a POC patch that implements "special values" for int GUCs.  We have quite a few GUCs
> with values that have special meaning atop other settings.  I have attempted to identify these and
> make it so you can specify a symbol name for these values instead of just relying on the magic
> number instead.
> 
> For instance, many GUCs have -1 for "disabled", so I've just made it so you can
> specify something like:
> 
>   SET log_min_duration_statement = disabled;
> 
> And the raw value will be set to -1 in this case.  For the purposes of testing, I have also added a
> new GUC "output_special_values" to affect whether `SHOW` or anything that relies on _ShowOption()
> can show with the special value instead of just the raw magic value, allowing tools to consume the
> original raw value, or provide the output to the user in the nicer format.
> 
> This has only been done for ints, and the passthru I did was very quick, so I have probably missed
> some options that didn't explicitly have their interpretations in the file and/or I didn't know
> about it already.  I do not think there are these sorts of values in other non-int GUCs, but there
> might be, so a similar approach could be taken to expand things to other config types in the future.
> 
> Let me know your thoughts; I personally find this to be useful, and would be a nicer way for some
> configs to be displayed in the postgresql.conf file.

As discussed on IRC, I am in favor of this improvement.  (I have not yet
looked at the patch.)
-- 
Vik Fearing