You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently, we have a completely inconsistent way of addressing uid / gid in our extfs scripts.
Here is an assortment of what we do in different cases:
Fake current user uid/gid
Fake current user uid, but set gid to 0
Fake uid/gid from file stat
Fake uid from file stat, set gid to 0
Always return 0/0
Sometimes numeric and sometimes symbolic names are returned.
This makes testing difficult and in particular next to impossible to make the test suite run as root.
My suggestion is to always return 0/0, if the FS in question does not have information about uid/gid, instead of doing 1.-4. Only if the FS does have this information, try to translate it appropriately.
My understanding is that mc will take current user uid/gid upon copyout anyways, so faking is not necessary. Going for 0/0 doesn't present any disadvantage I can think of.
Important
This issue was migrated from Trac:
zaytsev(@zyv)Currently, we have a completely inconsistent way of addressing uid / gid in our extfs scripts.
Here is an assortment of what we do in different cases:
Sometimes numeric and sometimes symbolic names are returned.
This makes testing difficult and in particular next to impossible to make the test suite run as root.
My suggestion is to always return 0/0, if the FS in question does not have information about uid/gid, instead of doing 1.-4. Only if the FS does have this information, try to translate it appropriately.
My understanding is that mc will take current user uid/gid upon copyout anyways, so faking is not necessary. Going for 0/0 doesn't present any disadvantage I can think of.