If I actually wanted to use a block language at my day job I'd want affordances for larger, more complex programs. Compiler or JIT, copy/paste, zooming in and out of parts of programs (ala Prezi or something), integration with source control diff tools, C extensions / integration with existing code, console output, file I/O, etc.
It could do absolutely everything that Lua could do. We also added a kind of multifile support, using tabbed views to reimplement libraries on the browser side of things (compiling into separate Lua files).
The idea was that educators at both ends of the spectrum could convert to and from the same visual representations. The primary school classes the uni were running could use blocks, and slowly move to normal Lua, one feature at a time, all the way up to the first and second year ICT programs for higher ed.
Because it could compile to Lua, we could use git and a whole bunch of other tools that already exist. There was an experimental C/asm.js branch to add in FFI support, but that was deemed too time-costly whilst the rest still had a bunch of edge-cases.
And then, unfortunately, the whole program got yanked. But about that time it was close approaching what I would consider "useable" for actual real work.