Hello Tom, 22.07.2019 7:14, Tom Lane wrote: >> Fixing both places sounds adapted to me. An alternative we could use >> here is just to say something like that: >> The effective resolution is only 1/HZ, which can be configured with >> kernel parameter (see man 7 time), and is 4 milliseconds by >> default. > Whatever we say here is going to be a lie on some platforms. > > Probably best just to say that the sleep resolution is platform-dependent > and leave it at that. I think, we can say "on many systems"/ "on most Unixen", and it would not be a lie. In my opinion, while a generic reference to platform-dependency is OK for developer' documentation, it makes the following passage in config.sgml vague (we don't give user a hint, what "the effective resolution" can be - several seconds/milliseconds/nanoseconds?): /The default value is 200 milliseconds (<literal>200ms</literal>). Note that on many systems, the// //effective resolution of sleep delays is 10 milliseconds; setting// //<varname>bgwriter_delay</varname> to a value that is not a multiple of 10// //might have the same results as setting it to the next higher multiple of 10. / -> /The default value is 200 milliseconds (<literal>200ms</literal>). Note that the// //effective resolution of sleep delays is paltform-dependent. setting// //<varname>bgwriter_delay</varname> to a value that is not a multiple of the effective resolution,/ /might have the same results as setting it to the next higher multiple./
Best regards, Alexander
