Re: [Chicken-users] openssl egg

[Thomas Christian Chust]

Shawn W. wrote:

> On May 23, 2007, at 6:38 AM, Thomas Christian Chust wrote:
>
>> [...]
>> your patch looks correct so far. The reason why I didn't include a
>> procedure like that in the egg in the first place is that its existence
>> suggests the possibility to wrap an SSL transport layer around any given
>> pair of input and output ports. That is unfortunately not the case,
>> though. Currently it is only possible to create an SSL wrapper for a
>> file descriptor.
>> [...]
> 
> You CAN make SSL wrappers for sockets, files, strings/memory, in C,
> using the bio API. Can they be wrapped up in scheme ports?
> [...]

Hello,

of course I know that it is theoretically possible to do both way
wrapping of OpenSSL I/O abstraction objects into ports and of ports into
OpenSSL I/O abstraction objects.

Actually implementing it in the general case, though, is a lot of
tedious work, extremely inefficient, cannot be done in a
SRFI-18-friendly way and is, in my humble opinion, very seldomly useful.
That, together with the reasons I already mentioned in my last posting,
made me refrain from implementing this scheme or some more restricted
functionality which would not only be restricted but also wouldn't be
very clean.

Now, if lots of people say that they desperately need this kind of
feature, I might add it to the egg despite my objections, but at the
moment it doesn't look like that...

cu,
Thomas