On 11/1/18 10:59 AM, Matt Johnson (AZURE) wrote:
The cited articles appear to be opinions and surveys only. I would like to see something, anything, from an official source
I'd also like to see something official. However, the Moroccan government appears to have made last week's announcement at the last minute to suppress political controversy that could have derailed the change, and it's not implausible that they'll do something similar for next year's Ramadan change - which would be a real problem for us and our downstream data users. When there's a conflict between "official" and "practical" time for past timestamps, I have preferred "practical" time on the grounds that users typically prefer the timestamps they actually used. This recently came up for El Paso, which would require its own Zone line if tzdata merely recorded official US time zone legislation; but tzdata doesn't do that, and instead reflects the timestamps that the vast majority of El Pasoans actually used during the period in question. Similarly, for Morocco we now have the situation where the government's officially announced plans do not currently include a Ramadan fallback. Here, too, we have a conflict between "official" and "practical" time, the main difference being that here the timestamps in question are predicted timestamps, not past timestamps. This is a judgment call, and I could certainly see the call going the other way in other circumstances. However, in this particular instance the case for including the guessed change is reasonably compelling: people on the ground are telling us that a Ramadan fallback is most likely what will happen, a poll has been published saying there's a broad consensus for a fallback, and Morocco has been doing such fallbacks for five years already. tzdata is always guessing about the future, and it's better to make the best guess we can even when the guess isn't official yet. (Besides, if we guess wrong now we can always fix it later. :-)