Jim Rees wrote:

> Don't you mean #(6 4 3 0) ?

Yes, there were bugs in one of my previous emails.  Those
tests were corrected before I submitted my pull request.

The pull request does contain some errors in comments for
select.scm, which you can find by running a diff against
Larceny's current version of that file.  Larceny's current
version of vmsort.scm also removes a gratuitous inefficiency
in vector-merge-sort, which can again be found by running
a diff between the two.

>  I also think that any new sample implementation code produced today
> should easily drop in to a conforming R7RS implementation and just work
> without changes.

Some renaming of files might be necessary because the R7RS
doesn't specify any mapping of library names to file names.
Since we already have that problem, I went ahead and put
two library definitions in a couple of files, which some
implementations of the R7RS probably won't allow.  Aside
from that, the code in my pull request should run in any
implementation of the R7RS.

Will