To answer my own question again:
The problem was solved when I realized that doing a select() on the socket file descriptor is not enough to know when there is data to read...
VERY IMPORTANT: You must also check SSL_pending(...) to see if there is data which openssl has readily available for reading.
A note about this in the documentation for SSL_read() would be nice (or at *least* a see also)
So it wasn't a hesitation to sent out data, but *my* hesitation to read it.
-- Davy
Davy Durham wrote:
I'm using openssl in some code that very much expects data to get sent when the write operation occurs.
I *think* I'm noticing openssl hesitating to write data sometimes. I'm not ruling out it being my doing yet, but when I remove openssl from the layers of code, I'm not seeing the problem.
Nagle's algorithm is disabled.
Could this be happening? (Not that I am but,) If I wanted to write 1 byte messages, should they get sent as I write them? If not, is there way to force this behavior? Is there a flush operation that I have to call? Is there a way to make it always do that?
I know data has to be written in records, but can a record be 1 byte?
______________________________________________________________________ OpenSSL Project http://www.openssl.org User Support Mailing List [EMAIL PROTECTED] Automated List Manager [EMAIL PROTECTED]
