Closed bovine closed 4 years ago
Thank you for finding this. I have also checked the other METARs from Australia and found this:
METAR YBBN 041130Z 15010KT 9999 SCT013 BKN028 BKN045 22/19 Q1018 FM1130 16011KT 9999 -DZ FEW010 BKN015 INTER 1130/1300 16020G30KT 3000 SHRA BKN010 TEMPO 1300/1430 2000 DZ BKN010 FM1130 MOD TURB BLW 5000FT TL1200=
Currently FMxxxx or TLxxxx or ATxxxx are expected to follow TEMPO or BECMG or INTER, and they are assigned Type::TEMPO or Type::BECMG or Type::INTER accordingly. Otherwise FMxxxx or TLxxxx or ATxxxx are assigned Type::NONE (when designing TrendGroup I assumed it does not happen in real life).
I agree on assigning Type::FROM to lonely FMxxxx group.
Regarding the rest, I think I’ll remove Type::NONE and add Type::UNTIL and Type::AT for lonely TLxxxx and ATxxxx groups.
Sequence FMxxxx TLxxxx not preceded by TEMPO or BECMG or INTER will be assigned Type::TIME_SPAN.
Also based on the METAR above, at the moment trends like TEMPO 1300/1430 or INTER 1130/1300 are not recognised, I would add them as well.
I can add all of this to version 5.0.0 and roll it out this week, what do you think?
ok sure, that release schedule is fine.
I may be able to identify more issues soon, as I am continuing to do more comparison/validity work this week.
Opening as a separate issue... Here's another case I found:
That gets called with:
CloudGroup FEW020 TrendGroup Type::NONE, FM0200 CloudGroup SCT030
I think the TrendGroup should have been called with Type::FROM instead? Do you agree?
Perhaps this block needs to set
result.t = Type::FROM;
https://github.com/nnaumenko/metaf/blob/59267179360103ba1822883fab059665c167c91f/include/metaf.hpp#L4098Originally posted by @bovine in https://github.com/nnaumenko/metaf/issues/1#issuecomment-586046077
Here's a condensed program for the above test case:
Originally posted by @bovine in https://github.com/nnaumenko/metaf/issues/1#issuecomment-586388155