Implications with "never" – seems like a bug

The Inform compiler will reject an assertion of the following form as too certain:

A thing can be fragile or sturdy. A thing can be red or blue.
Something blue is always fragile.

However, it will accept an assertion of the following form:

A thing can be fragile or sturdy. A thing can be red or blue.
Something blue is never sturdy.

However, the assertion appears to be ignored and have no effect.

But “never” is just the negation of “always”. If it’s going to reject an “always” implication, then it should also reject a “never” implication for the same reason.

Side note: I also saw a “seldom” implication that seemed to do nothing; however, I didn’t reproduce it in a minimal test case, so it may have been caused by something else interfering. I’ll report back if I figure that out.

1 Like

seldom is the opposite of usually. If you say X can be sweet or savory then X is usually savory or X is seldom sweet are no-ops 'cause they’re redundant.

re: the never thing, there are several bugs of the form “the compiler accepts this without comment but then does nothing”…