It seems to me that using the GUI to compact an hourly dataset to (7-day-) daily produces some hiccups. The example dataset is created with the small script from the related feature request ticket (link given below). Reminder: the range is 10:4 - 14:7, so incomplete first as well as last days.
- With the GUI/menu wizard, choose "use starting values". One would expect a new daily range of 10-14 or 11-14, depending on whether the convention is that on the first day (number 10) any first valid value is OK or not. However, instead the new range shown by gretl is 10-13 -- but notice that the variable "day" turns out to hold 11 through 14, so the displayed range numbers just seem to suffer from an off-by-one bug, possibly without any further consequences.
- Start over the experiment, but choose "use end values". The range now is 10-13 as expected. OK.
- Once again, but now taking averages. The "day" variable shows that the complete middle days 11-13 are retained, but gretl's displayed range is 10-12 instead. Again off-by-one?
BTW, notice that the compaction to daily is in principle supported in the GUI, contrary to the script interface - see the other request ticket filed because of that.
Related: https://sourceforge.net/p/gretl/feature-requests/228/
The off-by-one issue is now fixed in git, so I'll close this.