On Mon, Sep 2, 2024 at 5:47 AM Peter Smith <smithpb2...@gmail.com> wrote: > > Hi hackers. While reviewing another thread I had cause to look at the > docs for the pg_replication_slot 'inactive_since' field [1] for the > first time. > > I was confused by the description, which is as follows: > ---- > inactive_since timestamptz > The time since the slot has become inactive. NULL if the slot is > currently being used. > ---- > > Then I found the github history for the patch [2], and the > accompanying long thread discussion [3] about the renaming of that > field. I have no intention to re-open that can-of-worms, but OTOH I > feel the first sentence of the field description is wrong and needs > fixing. > > Specifically, IMO describing something as "The time since..." means > some amount of elapsed time since some occurrence, but that is not the > correct description for this timestamp field. > > This is not just a case of me being pedantic. For example, here is > what Chat-GPT had to say: > ---- > I asked: > What does "The time since the slot has become inactive." mean? > > ChatGPT said: > "The time since the slot has become inactive" refers to the duration > that has passed from the moment a specific slot (likely a database > replication slot or a similar entity) stopped being active. In other > words, it measures how much time has elapsed since the slot > transitioned from an active state to an inactive state. > > For example, if a slot became inactive 2 hours ago, "the time since > the slot has become inactive" would be 2 hours. > ---- > > To summarize, the current description wrongly describes the field as a > time duration: > "The time since the slot has become inactive." > > I suggest replacing it with: > "The slot has been inactive since this time." >
+1 for the change. If I had read the document without knowing about the patch, I too would have interpreted it as a duration. thanks Shveta
