migration hickup

[joerg van den hoff]

On 09.03.21 18:11, Ryan Schmidt wrote:
> 
> 
> On Mar 9, 2021, at 10:41, joerg van den hoff wrote:
> 
>> for the record and in case someone searches the mail archive: re-installing Macports from the .dmg installer solved the problem for me (no more warning messages regarding missing `macports' user).
>>
>> out of curiosity: any ideas, why this warning message is issued in the first place after the Migration Assistant run? maybe the issue is fixable on the Macports side for the future (I have seen this with 10.15.7)?
> 
> I don't think there's anything for us to fix really.
> 
> You reported that you used migration assistant and deliberately did not migrate the macports user. I concur that this was the correct choice, since if you had migrated the macports user, migration assistant would have relocated its home directory and you would then have had move the home directory from its relocated place to its correct place and used dscl to fix up the user's home directory entry.

yes, could be... as said, it sounded wrong to me doing that: agree to move the macports home dir to 
/Users. but probably that would have been the way to go, as you indicate.

> 
> Since you did not migrate the macports user, you should not have had a macports user or group on the new system. I'm not sure why some utilities seemed to indicate that this user existed. I agree with Rainer that the solution was to rerun the MacPorts installer so that it would create the user and group and the 

indeed, that is the only thing which I don't get: `finger', `id', `dscl' all agreed that the 
macports user still existed (and its home dir was still in its canonical place in the /opt/local 
tree rather than moved somewhere else, of course).

so I find this strange: seemingly everything is just fine (macports user and its home dir exists 
etc.) but I get that "cannot change owner: no macports user" warning nonetheless. so this looks like 
OSX had not a consistent view of the state of affairs across different utilities at that point? 
notably `dscl' knowing that user is not sufficient? what, then, is the reliable/correct way to 
decide whether a user account "really" exists? but whatever ...

user's home directory.
> 

understood. will try to recall the procedure in due time when doing another migration to a new 
machine...

best,
joerg