On Fri, Sep 26, 2014 at 2:39 PM, Stephen Frost [via PostgreSQL] < ml-node+s1045698n5820714...@n5.nabble.com> wrote:
> David, > > * David Johnston ([hidden email] > <http://user/SendEmail.jtp?type=node&node=5820714&i=0>) wrote: > > This is 9.5 material because 1) it isn't all that critical and, 2) we > DO > > NOT want a system to not come up because of a GUC paring error after a > > minor-release update. > > Agreed. > > > I don't get where we "need" to do anything else besides that...the > whole > > "actual min values" comment is unclear to me. > > Well, for cases that allow going to zero as an "off" option, we've > already decided, I believe, that sub-1-unit options are off the table > and so the min value is at *least* 1, but there could be cases where '1' > doesn't actually make any sense and it should be higher than that. > > Consider the log file rotation bit. If it was in seconds, would it > actually make sense to support actually doing a rotation *every second*? > > No. > > In that case, perhaps we'd set the minimum to '60s', even though > technically we could represent less than that, it's not sensible to do > so. The point of having minimum (and maximum..) values is that typos > and other mistakes happen and we want the user to realize they've made a > mistake. > > What needs to happen next is a review of all the GUCs which allow going > to zero and which treat zero as a special value, and consider what the > *actual* minimum value for those should be (excluding zero). I was > hoping you might be interested in doing that... :D > > Like I said I just want to fix the bug and call it a day :) For me just enforcing 1 as the minimum for everything would be fine. I'd rather mandate non-integer data entry than impose an actual minimum that is greater than 1. Specifically a too-short/too-small value might be used during exploration and testing by a new user even if the same value would never be useful in production. That, in fact, is the one reason that allowing "5s" for log rotation age would make sense - to allow people to more easily checking their log rotation policies. But making it work without disrupting people using "=60' (1hr) is impossible without simply outlawing unitless values. David J. -- View this message in context: http://postgresql.1045698.n5.nabble.com/proposal-rounding-up-time-value-less-than-its-unit-tp5811102p5820717.html Sent from the PostgreSQL - hackers mailing list archive at Nabble.com.