[TUHS] History of chown semantics

Clem Cole clemc at ccc.com
Sat Jan 11 01:16:06 AEST 2014


On Thu, Jan 9, 2014 at 8:01 PM, Larry McVoy <lm at bitmover.com> wrote:

> On Thu, Jan 09, 2014 at 07:15:55PM -0500, Brian S Walden wrote:
> > Now you are going to say this could all be done with proper use of group
> ids
> > and group permissions.  I agree, but in practice it was not done
>
> Bzzt.  We have a solution, they should have used it


Amen.

Brian - thank you for explaining from where the feature came.  It makes
more sense to be that issue that cause it was clearly the small number of
UID the incorrect use of GIDs and groups.   A fix to UID was later
implemented in the other editions, and I expect if PWB had done it earlier
- dmr et al would have "bought back" that one also.

While I'm glad to hear the explanation.  I admit that do not buy the IBM
programmer stuff. I know for a fair  number of us, from friend wnj, the
late Ted Kowalski (aka frodo) to many of us others programmed on TSS, MTS
as well as other IBM systems before UNIX.   The concept of file give-away
still seems far fetched as an explanation of it helped the user base.

Greg's example if his wife were in his group and it was properly used, its
not an issue (all of my family are included in group "ccc" on our all
systems at home and I do exactly the same thing).    Brian said - "proper
use of group ids." -- I think its going to be hard to convince a number of
us that file give-away is a good idea when you have groups.  Frankly, I
think it violated Mash's own "small" change directive - as we have together
pointed out it has so many side effects. The times where it actually might
be helpful seems pretty small and all of those case so far that we can
think of there are better solutions already baked into UNIX.  The negatives
of a file give-away feature (at least to me and I suspect many if us who
find it distasteful) are so many.

All that said -- the mystery and original Tim question that started this
thread has been answered...  thank you all.

Source -- PWB (1.0)
Why - seems to be it solved an issue with how UID/GIDS were used in
practice in the PWB world.

IMO:  it's to bad it got baked in SVID and POSIX when we had other
solutions the problems that needed to be solved.  But clearly others such
as Brian and Greg like it.  Surely there is no accounting for taste ;-)

Clem
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20140110/aaa12335/attachment.html>


More information about the TUHS mailing list