metering issueshttps://gitlab.common-lisp.net/dkochmanski/metering/-/issues2019-10-26T00:36:40Zhttps://gitlab.common-lisp.net/dkochmanski/metering/-/issues/3Add processor cycles to metrics2019-10-26T00:36:40ZSteveAdd processor cycles to metricsThe time macro, at least on SBCL, returns the CPU cycles. That metric is a useful proxy for overall performance. Probably best to return nil for those implementations that are unable to obtain this value.The time macro, at least on SBCL, returns the CPU cycles. That metric is a useful proxy for overall performance. Probably best to return nil for those implementations that are unable to obtain this value.https://gitlab.common-lisp.net/dkochmanski/metering/-/issues/2Feature: Return raw profiling data2019-10-29T07:24:02ZSteveFeature: Return raw profiling dataRaw profiling data is important to determine if the means have been affected by outliers, for plotting, and to see if there are any correlations with the environment, implementation or OS on performance.Raw profiling data is important to determine if the means have been affected by outliers, for plotting, and to see if there are any correlations with the environment, implementation or OS on performance.https://gitlab.common-lisp.net/dkochmanski/metering/-/issues/1Support overriding get-time (or gettimeofday)2018-08-25T21:41:51ZMichael RaskinSupport overriding get-time (or gettimeofday)Currently `get-time` uses `get-internal-run-time` (with millisecond resolution in most cases) and doesn't support overriding this choice.
I would like to be able to sometimes use `CFFI` and its wrapper around `gettimeofday` syscall, w...Currently `get-time` uses `get-internal-run-time` (with millisecond resolution in most cases) and doesn't support overriding this choice.
I would like to be able to sometimes use `CFFI` and its wrapper around `gettimeofday` syscall, which often provides microsecond resolution.
Is supporting overrides or conditionally supporting `CFFI` a better way to do this? I can prepare a merge request for either approach.