PROPOSED Are we happy with boolean header arg semantics?

  • State "PROPOSED" from "" 2010-02-24 Wed 15:51

For boolean header args like cache, noweb and colnames, the value is <on> if the header arg is yes; absent or any other value is <off>.

Is it worth considering nil or absent = <off>, and any non-absent value = <on>, in order to be in line with lisp?

I think this would also make sense from the point of view of least-surprise, because the default sense of those args is <off>; and anyone who is supplying them is likely to be wanting to turn them <on>, and so the only way for them to fail to do what they want would be if they supplied nil.

In particular, users might be surprised that t, y, Y, on, Yes don't work.

I guess we're going to have to really want to make this change to accept the backwards incompatibility cost, but I'm just noting it down. It could always be implemented by a variable.