I suspect that there is a set0.def which could be imported, and this
would give the Rochester set0 as they intended. In particular, it has a
non-standard set header. The information of a problem set is more than
the list of files, so I wouldn't expect a directory listing from the
Library Browser to reproduce previous sets exactly.
Here, when scanning the directory, it picked up a set header file (or
two). It also probably missed a problem which is normally part of set 0
because it was in a subdirectory with graphics. I guess the upshot is
that you can't rely on files found in a directory to be a usable set.
I can see several potential changes which might help. One would be for
the Library Browser to skip all header files. Actually, I thought it
already did that, but maybe I am mistaken (or maybe that was added in a
more recent version). The downside of skipping them, is that then it is
hard to preview any header files just sitting on disk.
Another change would be for it to pick up subdirectories if they have a
single .pg file in them, and treat them as if they are in the parent
directory. I expect a future version of the Library Browser to do this
thanks to Davide Cervone.
<| Post or View Comments |>