Greg Allen

+1

Actually, since `createuser` finds by username, the first case isn't necessarily a problem. It will update the user without changing the `uid`. The `uid` that was provided is simply ignored.

Agreed, and it might be nice to know that you didn't get the `uid` that you specified. Here are the two conditions: ``` # uid already exists, but username does...

I think echoing the error and `exit -1` is appropriate for a CLI install, but not very friendly for a GUI install. I could `osascript` a dialog if I knew...

From `man installer`: ``` ENVIRONMENT COMMAND_LINE_INSTALL Set when performing an installation using the installer command. ```

Even better: `createuser` is already reporting and returning an error. It's just not getting checked.

Unfortuately, `createuser` leaves a broken user when it fails. The above tests are better, unless we make changes to `createuser`.

The error handing in [createuser](https://github.com/gregneagle/pycreateuserpkg/blob/c2ee28af902e6354937bed69d7aca2751c6d7554/createuser/createuser/main.m#L65) has a problem. In `createUserRecord` when `createRecordWithRecordType` fails a message is reported, but the erroneous record is still returned. Then in `main`, `setAttributesForUser` gets called...

@robertgendler said on #90 >As the PR currently exists, it needs an absolute path when fed the option --pwpolicy_file, please re-submit and account for relative paths. The PR's behavior is...

Yubikeys are now a requirement at our lab, and new Macs come with Catalina. So this issue is affecting our ability to purchase Apple hardware. Can you please point me...