It turns out that lots of scripts still run zwrite without setting any locale environment variables. The charset of the default C locale is ANSI_X3.4-1968 (ASCII), so we were sending z_charset = ZCHARSET_ISO_8859_1. But UTF-8 is much more common these days, so this results in a lot of mislabeled zephyrs.
Other clients just ignore z_charset (Roost made a good-faith effort to respect it, and gave up after finding that it really has no correlation with the actual charset); but it still caused incorrect display in zwgc.
Since UTF-8 is just as good a superset of ASCII as ISO-8859-1 is, we should just assume UTF-8 by default in this case.
(We still assume ISO-8859-1 if the locale explicitly specifies that, such as en_US. Almost nobody uses such locales anymore.)
davidben
commented
10 years ago
It turns out that lots of scripts still run
zwritewithout setting any locale environment variables. The charset of the default C locale is ANSI_X3.4-1968 (ASCII), so we were sendingz_charset = ZCHARSET_ISO_8859_1. But UTF-8 is much more common these days, so this results in a lot of mislabeled zephyrs.Other clients just ignore
z_charset(Roost made a good-faith effort to respect it, and gave up after finding that it really has no correlation with the actual charset); but it still caused incorrect display inzwgc.Since UTF-8 is just as good a superset of ASCII as ISO-8859-1 is, we should just assume UTF-8 by default in this case.
(We still assume ISO-8859-1 if the locale explicitly specifies that, such as
en_US. Almost nobody uses such locales anymore.)