17 Jun
2001
17 Jun
'01
7:01 a.m.
Alexander Bochmann <ab(a)gxis.de> wrote:
I have the "Managing uucp..." book here, and
"Using UUCP..."
at work... Unfortunately, the books seems to be neither on
one of the ORA Bookshelf CDs, nor on Safari...
The sad bit is, I know I have this book, just can't remember where I
left it last.
From a quick scan through the book, it seems
you're right.
But if you don't want to use uucpd, probably only a small
perl hack is enough to replace it's basic functionality
and possibly fake a suitable environment...
Well, it would be easy enough to hack different authentication into
uucpd too, and make the hacked uucp setuid the user you want to use
for uucp. Would that be sufficient though? I think Taylor UUCP gets
away with doing its own authentication and running all peer requests
as the uucp user because it does not rely exclusively on Unix
filesystem access controls for validaction of the peer's requests.
I'm not sure how true that is of HDB UUCP.
I assume that Taylor UUCP is old-style enough...
If I had to run a non-leaf system, I would chose to install
Taylor UUCP regardless of possible lack of coolness ;)
I remember that there used to be protocol/performance advantages to
Taylor UUCP vs. other UUCP implementations. I don't remember the
details, but they were sufficient to make the (small) trouble of
building it for SunOS 4.1.1 worthwhile.
Just looking at the docs, though, if I were building something to
support UUCPs running on various oddball classic iron peers I might go
for Taylor UUCP because it does seem to support some protocols that
are specific to non-Unix UUCPs ('y' and 'v' protocols especially).
-Frank McConnell