[imp] IE "page cannot be displayed"

[Eric Rostetter]

Quoting Lord Apollyon <implist at paypc.com>:

> It's a pity you couldn't or didn't have the time to see exactly what the SSH
> negotiation problems were.  3 years ago, hrm... Was this OpenSSL 0.9.5?

Yes, it is a pity.  But when you have a critical service, you sometimes have
to just fix the problem asap, rather than diagnose it.

Not sure what OpenSSL it was off hand.  Something on Solaris 7 is all
I remember...  I'm not able to look it up at this time.  I know right now
I'm running 0.9.6 on the machine, but that may have been updated since the
original roll out.

> Bizarre browser interop (SSL) issues typically stem from no shared cipher,
> to RSA keysizes being too large (or too small) - which is why even though
> you can use larger server keys, 1024 bits is best for compatibility.

I'm sure it isn't key size.  But beyond that, no idea.

> Do you recall exactly which browsers had the problems?  Even one or two of
> them?  I have Macintoshes and Wintendo boxes.

Nope.  No idea.  This was back when we had everything from IE 3 to IE 5,
Netscape 2 to Netscape 4, etc.  Running on MacOS 7-8, Windows 95-98 or NT4,
etc.  Lots of old variables that probably don't exist much any more, but
I'm too afraid to remove that code just in case they still do exist.

Now, on my Horde server I don't do that, I've gotten no ssl complaints related
to this (though I've got lots of *other* complaints caused by broken browsers
but no ssl related ones).  But then my Horde user base a rather homogeneous
group of only a few thousand, rather than a diverse group of almost 30K and
growing.

My experience is: the more deverse the user base, the more such problems
arise.  So if possible keep your user base small ;0

> I'm curious now.

Yeah, sorry I don't have more details.  Just don't have the time.

> =Apollyon=

--
Eric Rostetter
The Department of Physics
The University of Texas at Austin

Why get even? Get odd!