Stephen Frost <sfrost@snowman.net> writes: > I understand that there may be objections to that on the basis that it's > work that's (other than for this case) basically useless,
Got it in one.
I'm also not terribly happy about leaving security-relevant data sitting around in backend memory 100% of the time. We have had bugs that exposed backend memory contents for reading without also granting the ability to execute arbitrary code, so I think doing this does represent a quantifiable decrease in the security of pg_hba.conf.
The Stephen's proposal changes nothing in security. These data are in memory now. The only one difference is, so these data will be fresh.