Autor: Ian FREISLICH Data: A: Chris Lightfoot CC: exim-users, Jeremy Harris Assumpte: Re: [exim] data-phase efficiency
Chris Lightfoot wrote: > On Wed, May 25, 2005 at 09:43:00AM +0100, Philip Hazel wrote:
>
> > Fascinating, but I'm afraid that's all completely over my head! I'm
> > really not good at the SSL stuff, and remember, Exim supports GnuTLS
> > as well as OpenSSL. What is currently there seems to work, and as I
> > have far too much other stuff to do, my judgement at the moment is
> > to leave well alone.
>
> it's nasty, because it can make a select loop much more tangled. One
> alternative would be to fork a process to proxy between SSL and
> non-SSL connections, though this has its own disadvantages.
Indeed, although I do prefer a more tangled select loop. Then, if
you remember how this thread started that suggestion is not even
worth contemplating :) The overhead of alarm() pales into
insignificance next to the overhead of fork().
On a side note, has anyone ever profiled the exim code?
