[Savannah-hackers] https port on subversions

[Joel N. Weber II]

For at least the second time today, I pointed my web browser at
https://subversions, and spent far too long being confused about why
it wasn't working.

(Which is to say, I'd made the same mistake at some point in the
distant past.)

I tend to think that if there's an end user operation that *I* can't
get right, then something is hopelessly badly designed.

It's also the case that subversions has more IP addresses available at
this point than it used to.  At VA, getting another IP address was a
problem.  At Global NAPS, subversions is sitting on a subnet that has
addresses for 14 hosts/routers, of which one is used for a Global NAPS
router, one is service-entrance.gnu.org, and the remaining 12 are
available for subversions and any servers we might add in the future.
If we had some reasonable reason to want to use 6 or 8 IP addresses on
subversions, we could do that and it wouldn't be a problem.  (We
don't, however, have enough IP addresses for one IP address per
package, and Barton is likely to question us if we request such a
thing.)

Anyway, I'd like to propose a few changes:

1) Remove pserver from the HTTPS port associated with the
``subversions'' IP address.  We should either run the same https
daemon that we run on ``savannah'', or just have that port refuse
connections.

2) Make sure that the HTTPS port on subversions-ssh.gnu.org is
listening for ssh connections.

I think it's the case that providing ssh instead of pserver is the
right thing; if people feel that they have to use pserver, I'd be
curious to know why.