timedelta marks time in days, seconds, and microseconds. It doesn't take leap years into account because the concept of years is irrelevant to timedelta. If you need to account for leap years, you need a different API.

[–] nilloc@discuss.tchncs.de 1 points 5 hours ago

365.25*10 would at least get you closer.

[–] Randelung@lemmy.world 1 points 15 hours ago

You can subtract two dates and get the exact time difference.

[–] wols@lemmy.zip 11 points 1 day ago (2 children)

The comparison is somewhat awkward, because the rails example presumably produces a date, while the python one is referring to an interval of time.
Just from the meme it's not obvious which was the actual intended use, so labeling either as inaccurate requires us to make assumptions.

Personally, the concept of "10 years ago" is a bit nebulous to me. If today is February 29th, is ten years ago March 1st? Doesn't seem right. Or particularly useful.

[–] nilloc@discuss.tchncs.de 1 points 5 hours ago

Ruby should add 10.years.ago.today

[–] eager_eagle@lemmy.world 4 points 1 day ago

yeah, that's pretty much why timedelta doesn't have the concept of months or years, just days and smaller units. I like it better this way.

[–] eager_eagle@lemmy.world 8 points 1 day ago* (last edited 1 day ago)

Because what's accurate here depends on the context, and the Python example doesn't hide that from the programmer.

The same dilemma goes for month calculations: does "3 months ago" mean 90 days ago, 91.3 days ago, this many days into the target month, or this many days from the target month's end (e.g. to account for 28, 29, 30, and 31-day months)?

[–] Blackmist@feddit.uk 3 points 1 day ago

Does the Ruby version do that though?

I haven't got it installed to check, but seeing constants like SECONDS_PER_YEAR in the documentation makes me think it's just as bad if not worse.

https://api.rubyonrails.org/classes/ActiveSupport/Duration.html