This PR needs unittests, but i'm still thinking how to do them.
But the code was tested and works.

@ned21 feel free to review the code. i'll try to add unittests next week

@ned21 please rereview. tests fail, will be fixed later

tests are complete

@jrha i'm moving the milestone. we are seeing something odd, and at least the reporting needs refinement.

@stdweird any update on the strange behaviour?

@stdweird any update? Still seeing strangeness or is this now understood and ready for merging?

@jrha i didn't find the time yet to debug further. i moved the milestone.

👍

@stdweird got any further?

not really. sometimes, for whatever reason, the current GID is a single group; and in that case something goes wrong try to restore the group (looks like you can reduce the number of effective groups one is in)
eg. here the starting GID is 0 (normally it's a more than one group), and the resulting group

2018/06/05-10:24:23 [VERB] Got user id 163 (gid 163) (is_id 0 is_user 1)
2018/06/05-10:24:23 [VERB] Changing EGID from '0' to '163 163' with GID '0'
2018/06/05-10:24:23 [VERB] Changing EUID from '0' to 163 with UID 0
2018/06/05-10:24:25 [VERB] Restoring EUID from '163' to 0 with UID 0
2018/06/05-10:24:25 [ERROR] Something went wrong restoring EGID from '163 163' to '0' with GID '0 163': new EGID '0 163', no reason given

Hmm, odd.

Bumping this again, will see if I can take some time to look it more closely.

Rebased.