flow_view() chokes on reexports

[moodymudskipper]

And also chokes when using a binding in the global env :

flow::flow_view_deps(tibble::lst)

flow::flow_view_deps(dplyr::lst)
#> Error in getNamespace(ns): invalid data of mode 'character' (too short)
a <- tibble::lst
flow::flow_view_deps(a)
#> Warning in read.dcf(system.file("DESCRIPTION", package = pkg)): cannot open
#> compressed file '', probable reason 'No such file or directory'
#> Error in read.dcf(system.file("DESCRIPTION", package = pkg)): cannot open the connection

It's not very clear to me what's expected there, I guess it's not too bad if we provide the same diagram for flow::flow_view_deps(dplyr::lst), but it gives the impression that the diagram shows what happens in dplyr.

In the latter example it might give the impression that a is a tibble function.

We could also just fail for those, but that's depriving the user of some info we have at hand.

Maybe a warning, I dislike them but it's most probably used interactively anyway.

Have a reread some time from now, current opinion is : make 2nd case identical to 1st, 3rd case same with a label, give warning for reexports/non matching bindings.