Igor Sysoev ha scritto:
[...]
So, please let me know if I'm right:
- when a call to ngx_http_output_filter with a buffer chain of *only
one* buffer returns with NGX_OK, it means that the *entire* buffer
has been sent to the client.
Yes. Or if gzipping is enabled the entire buffer is buffered inside zlib.
If nginx return NGX_AGAIN then I have to set the
request->write_event_handler so that my handler is executed when the
socket is ready to send data again.
Yes.
I have done some tests with mod_wsgi.
If a WSGI application returns a big buffer, ngx_http_output_filter
returns NGX_AGAIN, *however* the entire buffer is sent to the client.
This is the same behaviour with a "full" buffer chain, with the
difference that even if I return NGX_AGAIN from my request handler, the
client reads the full response (with a "full" buffer chain in a previous
test, I have to return NGX_OK, when receiving NGX_AGAIN).
So, it seems that nginx will send the entire content of a buffer, as
soon as control returns to it.
This is important for me: when ngx_http_output_filter returns NGX_AGAIN,
should I have to send again the previous buffer when my handler is
called from the write event, or I can safely send the next buffer?
15 messages in ru.sysoev.nginxRe: sending data in "chunks"
