> with this patch, you can now restore default value with SIGHUP when
> commenting an variable in postgresql.conf

This seems pretty poorly thought out, in particular making PGC_S_SIGHUP
>= PGC_S_OVERRIDE seems to me likely to break the interaction with other
sources.  Doesn't that cause postgresql.conf to override per-user and
per-database settings?  Why do you need the extra value at all ---
isn't the correct logic just to reset entries with source PGC_S_FILE?

Another problem is that if there's something wrong with the config file,
this will cause all values previously read from the config file to
revert to defaults, which seems less than robust to me.  You really
shouldn't apply the reset until after the file has been parsed.

                        regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 5: don't forget to increase your free space map settings

Reply via email to