Indeed, that seems to be the way forward, although even that wasn't simple.

I tried

printenv | grep en_US

which returned

LANGUAGE=en_US

so I tried changing the LANGUAGE variable to en_US.utf8, en_US.UTF-8 but still got this error

abort: decoding near 'Sébastien ': 'ascii' codec can't decode byte 0xc3 in position 1: ordinal not in range(128)!

Eventually, I found using the LANG variable would work

setenv LANG en_US.utf8

You might need to pass that tip along to jwe or anyone else who has this issue.

@Sébastien: Of course, I respect people's names.  In this case, Mercurial refused to allow the changeset to be committed until I changed that character.  It was either make the change or give up on the patch altogether.  Perhaps Mercurial was less strict about file encoding in the past, or other reviewers know a way of bypassing these checks that I do not.

Small typo - the first bug fix release will be 5.2.0.

There is a bug in the code for ordeig that I contributed. The trivial 1x1 matrices are not handled and lead to a crash:

octave:1> ordeig(1)
error: tril: requested diagonal out of range
error: called from
    ordeig>isquasitri at line 105 column 10
    ordeig at line 50 column 7

while it should simply return 1.

The attached patch fixes the issue and adds a regression test case.