Ihor Radchenko writes: > Ping ;) So I gave up on this specific patch because I wrote a patch to just rewrite the entire `org-clock-sum' function using org-element API. Attached is the `org-clock-sum' rewrite patch which I've been using for a while with no issues. I have half finished patches locally to add more clocktable tests and to add clocktable benchmarks which is why I hadn't submitted this yet. This probably belongs in this email thread instead: https://list.orgmode.org/87y18vxgjs.fsf@localhost/ I believe I fixed all the points you brought up in that thread. While this patch is probably ready to merge, doing so might cause regressions to the fix applied in commit fd8ddf2874ca00505aa096c6172ea750cd5e9eaa. Ideally the fix in that commit should be ported to the org-element API. Notably, the malformed clock from the email thread from that commit is parsed a little strangely by org-element. I'm not sure what effect this has on my rewrite patch but regardless, we should probably fix this. Notice how ":day-end" and ":minute-end" are set but not ":hour-start" or ":minute-start". I have attached a rough patch adding a test for this case but my brain is currently melting from a heatwave so it might take me a while to make it into something good. Feel free to work on this yourself if you have time. "CLOCK: [2012-01-01 sun. 00rr:01]--[2012-01-01 sun. 00:02] => 0:01" (clock (:standard-properties ... :status closed :value (timestamp (:standard-properties ... :type inactive-range :range-type daterange :raw-value "[2012-01-01 sun. 00rr:01]--[2012-01-01 sun. 00:02]" :year-start 2012 :month-start 1 :day-start 1 :hour-start nil :minute-start nil :year-end 2012 :month-end 1 :day-end 1 :hour-end 0 :minute-end 2)) :duration "0:01"))