tags 379117 help thanks Jamm!n Wheeler <[EMAIL PROTECTED]> writes:
> Package: gtimer > Version: 1.1.6-11 > Severity: normal > gtimer's concept of "today" begins strictly at midnight. This is > nonsensical for many hackers, whose day is perhaps halfway through at > that point. There is a -midnight command line flag that lets you change the time at which midnight starts which partly remedies this. > It would be considerably more useful if, instead of a "Today" timer, the > left-hand timer was simply a sub-total whose start and finish was > decided by deliberate user action. For instance, I currently have to > report my time monthly (but *not* calendar monthly, so gtimer's reports > are not helpful), so it would be most useful to me if I could see the > month's total on the left, while keeping the overall total on the right. > Anyway, that's wishlist. This is the bug: The automatic rollover at > midnight causes a problem if you have left gtimer idling. Clicking > revert or resume only removes the time elapsed "today", it doesn't > rewind back to when the idle actually started if midnight has > intervened. Since it's also not possible to remove elapsed time from > yesterday using ^D etc, this permanently mucks up the report. > So I think the wish and the bug are tied together. If the left-hand > timer's lifespan were user controllable, there would be no automatic > rollover so the bug would not present. Yup, as soon as gtimer hits midnight (offset or otherwise), it records all the times into totals and creates new daily task entries, and after that you can't change the totals from inside the application. Unfortunately, this is pretty deeply entangled in the data structures and I don't see a straightforward way of fixing this, even to just fix the bug without addressing the wishlist item. Basically, when you recover from an idle, you'd need to walk backwards through task entries until you reached the beginning of the idle, subtracting off time as appropriate, but gtimer internally doesn't have the data structures that let one easily do that. I'm happy to apply patches, but I'm afraid that I'm not likely to have time to implement this myself in the near future. -- Russ Allbery ([EMAIL PROTECTED]) <http://www.eyrie.org/~eagle/> -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
