The radtest script parses the ppphint correctly. A trival test verifies that. Can you be more specific as to what the problem is?

I have problem with these two cases:
1) /usr/bin/radtest -x -4 testuser RedHat localhost:1812 1812 RedHat -1 localhost | grep -q 'PPP'
2) /usr/bin/radtest -x -4 testuser RedHat localhost:1812 1812 RedHat YES localhost | grep -q 'PPP'

Return code 1 is expected for these cases in our tests. Currently these commands return 0.

radtest was patched by us to handle the PPPhint, our logic said truth was any positive integer. Upstream did not implement the same logic when then fixed radtest, instead upstream implemented truth as any argument which is non-zero. This is why the last two tests were failing, it was expecting -1 and YES to disable the PPPhint but because the test is anything which is non-zero these arguments enable PPPhint. This behaviour changed I believe with freeradius 2.2.3 (not positive).

There were also several problems with the script which I had to fix in addition.

* The shared secret was not updated in the radius config to RedHat, the default is testing123. Therefore the shared secret of RedHat was changed to testing123.

* The users file was not updated with the test user and password. Now the users file is backed up and a user named "testuser" with a password of "RedHat" is created as per the test expectations.

* The radius server was not started nor was it restarted after updating the users file with the testuer, now the daemon is managed in the script (using the newly introduced RadiusdStart and RadiusdStop functions.

Closing this as NOTABUG because we'll follow upstream and all the issues are really in the test script. I will attach the version of runtest I updated to bug #1021541 where we're tracking the test updated for FreeRADIUS 3.