[Lcdproc] Does anyone on the list use the sample METAR client?

[Ethan Dicks]

I''ve been using lcdmetar.pl on and off for a while and finally
started to dig into some issues with unanticipated values in the METAR
returned for certain non-US situations.  The issues are not with
lcdmetar.pl per se - they are with Geo::METAR.  I've written Jeremy
Zawodny, the author, but I have not received a response.  Given that
the code is many years old, I'm not surprised.

The fundamental issues I'm having aren't a problem with U.S. reporting
locations, so unless you are curious about the weather outside of the
U.S., you have probably never seen anything unusual.  In my case, the
two particular stations that give me grief are Williams Field, near
McMurdo Station, Antarctica (NZCM) and Amundsen-Scott South Pole
Station (NZSP).  In the case of NZCM, for starters, the station ident
doesn't start with a K, so 'NZCM' gets added to the 'unknown' list of
tokens.  Addionally, they don't use the US-standard statute miles
visibility, they use meters, so a number like '2400' ends up in the
unknown list.  Both of these should be issues for European and other
non-US users of lcdmetar.pl/Geo::METAR.  On top of those, the METAR
for NZSP routinely writes the temp as 'M39/' without the trailing
dew-point/frost-point temp, a valid, but uncommon way to write a METAR
(the humidity and temps being such that in the winter, at least, the
Met guys don't calculate the dew-point temp).

I've tweaked up Geo/METAR.pm a bit, but wanted to see if anyone else
was having issues with non-US locations so that we could perhaps
compare our efforts.  If anyone is simply using Geo::METAR with
LCDproc and has seen odd 'undef' warnings or strange blank fields on
their LCD, please feel free to send me specific METAR strings that
don't parse well, and I'll see how they do run through my modified
METAR.pl.

Thanks for any feedback,

-ethan