Avoid username truncation with 'stat'

[bibliocar]

'ls' truncates usernames containing more than eight characters, making "bibliocar" into "biblioca". I found 'stat' worked in my kiss install without any problems in place of it. I rewrote both functions that used 'ls' to make use of 'stat', instead.

While doing so, I was unsure why: equ "$LOGNAME/$user" "$user/$LOGNAME"

wasn't: equ "$user" "$LOGNAME"

[E5ten]

stat was already used in the past, it was changed to ls because stat is not POSIX (and the flag to choose the output format is inconsistent across systems). Maybe a way to avoid this issue would be to add the -n flag to ls, which rather than outputting user names outputs uids, and that can then be compared to the output of something like id -u, rather than $LOGNAME?

[dylanaraps]

POSIX states the maximum length of a login name to be 8 (plus NULL terminator).

{_POSIX_LOGIN_NAME_MAX}
    The size of the storage required for a login name, in bytes, including the terminating null.
    Value: 9

See: https://pubs.opengroup.org/onlinepubs/009696899/basedefs/limits.h.html

Not sure if this is even something worth fixing. Stat cannot be used given @E5ten's explanation above. Will think about it.

[dylanaraps]

Maybe a way to avoid this issue would be to add the -n flag to ls, which rather than outputting user names outputs uids, and that can then be compared to the output of something like id -u, rather than $LOGNAME?

I played around with this (good idea) but we still need the username in "string" form as the $user variable is given a value in this function (which is then used in myriad places by callers). ie, some method of converting the numerical ID to the username untruncated.

[bibliocar]

Oh, okay! I guess I'll just come up with a shorter username, then. Glad I submitted and learned something new about POSIX!

It's much nicer to see a username stream by in the output than a UID, whatever you end up doing.

Maybe the place to address it is during username creation?

[E5ten]

Not that I'm saying this needs fixing in light of that POSIX requirement, but if this was still desired, wouldn't the UID being equal to id -u guarantee that $LOGNAME would be the correct username, and then user=$LOGNAME could just be done after the comparison with id -u?

[E5ten]

If my assumption above about $LOGNAME being correct if the uid comparison succeeds is wrong, then alternatively a possibility would be doing the uid comparison with plain id output, with parsing probably going something like this:

output=$(LC_ALL=C id)
uid=${output#*=}
uid=${uid%%')'*}
user=${output#*'('}
user=${user%%')'*}

This is obviously very hacky though, and I don't really have reason to think my $LOGNAME is wrong yet, and if it isn't then this would be totally unnecessary and undesirable.

[E5ten]

Oh, and I forgot about the second conversion of ls to stat for the octal permissions. I think that part is just unnecessary, right? Like doesn't fix a potential problem or anything?

[LoganDark]

Thanks!! This patch saved my username of logandark :)

Now just to figure out how to keep the patch through updates...