Because the buffer is saved after Dante performs checks, the save confirmation appearing in the echo area can prevent the idle-induced type display from appearing. It's also possible for many requests to pile up if the idle-display threshold is set low, so the last valid idle type display is saved as a temp variable.
I'm not catching every failure case--sometimes haskell-doc.el will have its own type annotation already in the echo area and Dante will respect it; and shortly after startup, sometimes stuff won't idly display. I'm not sure how to resolve these cases, but this patch covers at least many of the other scenarios.
Because the buffer is saved after Dante performs checks, the save confirmation appearing in the echo area can prevent the idle-induced type display from appearing. It's also possible for many requests to pile up if the idle-display threshold is set low, so the last valid idle type display is saved as a temp variable.
I'm not catching every failure case--sometimes
haskell-doc.el
will have its own type annotation already in the echo area and Dante will respect it; and shortly after startup, sometimes stuff won't idly display. I'm not sure how to resolve these cases, but this patch covers at least many of the other scenarios.