[tz] Issue with Africa/Casablanca

[Paul Eggert]

David Gauchard wrote:
> Latest version of zic produces an empty TZ string with Africa/Casablanca (and El Aaiun)
> 
> Would it better be at least GMT0 ?

By "empty TZ string" I assume you mean the string at the end of the TZif file.

Traditionally, zic has omitted that string if it couldn't express far-future 
rules in the form of an extended POSIX TZ environment variable, as is the case 
for Morocco. Alternatively, I suppose zic could attempt to approximate 
far-future rules as best it can within the extended-POSIX framework, and use a 
string like "<+01>-1"; however, this could well give data consumers a false 
confidence in predicted timestamps. Perhaps it would be better to leave things be.