On Mon, Aug 21, 2000 at 04:58:03PM +0200, Nick De Roeck wrote:
>>> Check my original post (I'm copying it below). We're unhappy
>>> with OpenSSL's support for nonblocking I/O on servers; currently,
>>> it either wants you to rewrite your server to let OpenSSL handle
>>> the network I/O, or use BIO pairs to let OpenSSL think it's doing
>>> networking still. Neither alternative is appealing.
>> The TLS/SSL protocols are designed to run on top of a network
>> connection (surprise :-), so it's a rather obvious approach to have
>> the API designed accordingly.
> I'm currently writing a multi-platform webserver using openSSL and I do also
> experience problems with the fact that the networking code is too tighty
> integrated with openSSL. On a unix system this may seem like a non-issue,
> but my adventures trying to get openssl running for both NT, MacOS and
> MacOSX made me curse the networking layer quite often (f.i. there's no such
> thing as a BSD-style network calling convention on standard Mac (yes, there
> is one but I find this a wrong way to go, besides more code = more bugs)).
>
> If openSSL would be more of a library that runs on top af a stream (any
> stream, being network/stdin/whatever...) it will gain potential on more
> obscure/less known/non-unix platforms.
You can run it on top of any stream by either writing your own
BIO module, or by using a BIO pair (or a memory BIO).
I implemented an SSL/TLS extension for a web server written
in Common Lisp by using BIO pairs, and Common Lisp streams
certainly are different from Unix file handles.
______________________________________________________________________
OpenSSL Project http://www.openssl.org
Development Mailing List [EMAIL PROTECTED]
Automated List Manager [EMAIL PROTECTED]
