Bug #43666 for Geo-METAR: Missing Temperature Bug

Sir: Partly, I am writing to thank you for already fixing the bug we were experiencing. At present, we are running version 1.14 of Geo::METAR, and I just inherited the job of finding out why some Canadian stations had temperatures around zero Fahrenheit for hours, then suddenly 32.0, then back to zero-ish. It turned out that the bad METAR obs had a temperature but no dewpoint, and your code was insisting that stations must be reporting both temperature and dewpoint. But I see that version 1.15 has changed that line to read like this: elsif (($parsestate >= $expect_modifier) and ($parsestate < $expect_pressure) and ($tok =~ /^(M?\d\d)\/(M?\d{0,2})/i)) Clearly, that problem has now been fixed. Thank you. As soon as my sysadmin gets in today, I plan to have him update our Geo::METAR to the current release, and that will fix our immediate problem. But I wonder whether you might consider making the code just a bit more flexible. I don't know how likely this is, but it is certainly possible for a station to report a dewpoint but no temperature -- probably not likely, but possible. If that ever happened, your code appears likely to miss the dewpoint. But if you recoded the temperature parser the way you recoded the dewpoint parser, that should not happen. Yes, I recognize that such a change could create new problems. And, while I was an expert in reading METAR many years ago in the Air Force, I have forgotten much. I don't know for sure that changing the regex that way would not cause the parser to confuse this token with some other. I would have to defer to your expertise on that. But if it's possible to do this without causing new problems, please consider it. Thank you. Show quoted text