Hello Terrence -
thank you very much for your mail (sorry for the late reply but yesterday was a
rather busy day). As far as I know, APL has no distinct rank concept - J
introduced rank but the only thing in APL which can be used to apply functions
in an element wise fashion to the elements of nested structures is the so
called component operator (two dots in a row like the dots in a German "Umlaut"
character). Using this component operator you can implicitly iterate over
nested structures in APL. The re-implementation of 5 which Thomas currently
undertakes will extend the interpreter so that it will be possible to supply
additional information during a word definition which will allow words to be
recognized by the interpreter as being unary or binary. Such words will then be
applied in an element wise fashion to all elements of one or two corresponding
If one wants to avoid this element wise application, it is just necessary to
leave out the unary/binary distinction from the word definition - in this case
the interpreter will apply the word to the array as a whole (relying on the
fact that built-in operators which are unary or binary will be applied element
Concerning a source code repository: We are already using the SVN supplied by
SourceForge, so we should all rely on this as the central repository:
Currently we have no branches at all and the re-implementation is done in a
separate subdirectory in the main trunk.
Have a great day - sincerely, Bernd.