"Miles Sabin" <[EMAIL PROTECTED]> writes:
> I've seen Tony Rogvall's post on this topic from a few months ago,
>
> http://marc.theaimsgroup.com/?l=openssl-users&w=2&r=1&s=writev&q=b
>
> and I'd like to raise it again.
>
> I have excactly Tonys problem: I want to be able to accumulate several
> independent buffers worth of write data, HTTP headers and one or more
> buffers worth of payload (multiple payload buffers are especially
> likely in conjunction with HTTP/1.1 chunked transfer encoding), and
> write them out to be sent as a single SSL record _without_ there
> needing to be a copy to an intermediate buffer. I understand that
> the performance tradeoffs are different with SSL/TLS vs. plain TCP,
> but I still think this would be likely to be a win performance-wise.
I seem to remember that I commented on this in the original thread.
Just use buffered I/O.
I'd really be amazed if this produced any significant performance
improvement. If you think otherwise, it should be easy to demonstrate:
Profile your application using a copy to an intermediate buffer.
See what percent of the running time actually goes to the copy.
I predict it's infinitesmal.
Premature optimization is the root of all programming evil.
-Ekr
[Eric Rescorla [EMAIL PROTECTED]]
http://www.rtfm.com/
______________________________________________________________________
OpenSSL Project http://www.openssl.org
User Support Mailing List [EMAIL PROTECTED]
Automated List Manager [EMAIL PROTECTED]
