jsmorley wrote: ↑March 8th, 2020, 3:47 pm
I can't say that I entirely love the methodology that weather.com uses, as while I generally accept that the forecast temperature for "today" will be the high, and the forecast temperature for "tonight" will be the low, I don't think that is entirely set in stone. I expect that there is some possibility, no matter how remote, that a warm front could come sweeping in during the evening, and the "night" could actually be warmer than the "day" was. I think I would have preferred an approach where the forecasted "high" and "low" for both parts of the day was just always returned, no matter what the time of day. While not strictly logical, it would certainly be easier to deal with. Also, note that weather.com can't tell you what the "high" will be for "tonight", or what the "low" will be for "today". I could envision those being of some interest.
But, it is what it is. Once you wrap your head around the logic of it all, and decide how you want to deal with it, it all works pretty well.
I'm not exactly in love with that methodology as well, as I too was used to high/low and max/min from the TV and such, but I took the decision to distribute those temps into the daytime / nighttime sections ever since the wxdata days - although back then the temps belonged to the "whole" forecasted day. My skins are scroll based, meaning they don't show both daytime and nighttime conditions, so distributing the temps this way perfectly suited my design (otherwise, I would have had either the daytime or the nighttime part of my skins displaying 2 temps while the other was left with none).
Incidentally (and interestingly), nowadays weather.com directly distributes them the same way I do, at least internally, where one temp belongs to the "day" (or better said "daytime") section and the other to the "night" (i.e. "nighttime") one.
I guess they too figured out that, while not exactly "by the book", this distribution alleviated the "confusion" where you have dozens of conditions (e.g. wind, precip, UV index, icons, humidity and so on) specific to daytime and nighttime periods while the temps were the only ones excluded from this type of "structure". Of course, on the surface (i.e. on display on the main site) they are still labeled High / Low and Max / Min, but from a structural point of view this distribution makes just as much sense as the one displayed. As I already said, it isn't rocket science for the user to figure out which is which (i.e. high≈max≈daytime, low≈min≈nighttime)...