Re: Maybe deprecating libgweather

[George Barrett <bob bob131 so>]

On Mon, Aug 29, 2016 at 10:34 AM, Giovanni Campagna 
<scampa giovanni gmail com> wrote:
> Any comments / opinions?

Just yesterday I was fighting with gnome-clocks to try and insert a 
timezone instead of a city. Reading the libgweather source, I feel like 
the model it assumes is too rigid (+1 wrt cities requiring METAR 
codes). Worse than that, though, is the assumption by applications 
using libgweather that location is a preferred proxy for timezones. Of 
course, that's great for things like automatically setting DST, but 
setting just a straight time zone is outside the bounds of the model.

I tried writing a patch for Locations.xml to add some dummy entries for 
timezones, but such dummy entries would carry the extra baggage of a 
location like METAR codes and whatnot. It would be great if libgweather 
had a common API for querying locations and timezones in a way that 
still allowed transparent updates to DST in most cases.

So I'm +1 for deprecation as well. Maybe it could be reintroduced later 
after an API rework?