It's a little useless for the client to send it, but perhaps the server could take care not to close the connection just before (~second) of the client's announced close time, just in case the client was about to send something as the server would have otherwise closed.
Independently, and more usefully, the server could announce its keep-alive idle time (once it exists: #14204) and the client could respect it.
The text was updated successfully, but these errors were encountered:
Apparently this is seen in the wild only about 0.2% of the time and servers have stopped(?) using it. nginx might still send it? Apache used to and stopped?
11:40 < igrigorik> re Keep-Alive: it appears (at least based on HTTP Archive data) that ~17% of responses provide Keep-Alive (!!).. 5s is ~7%, 15s is ~2%.
More data here: https://discuss.httparchive.org/t/use-of-keep-alive-header-on-the-web/794
Now that we have
net/http.Transport.IdleConnTimeout
we could also advertise our timeout to the peer:https://tools.ietf.org/id/draft-thomson-hybi-http-timeout-01.html#rfc.section.2
It's a little useless for the client to send it, but perhaps the server could take care not to close the connection just before (~second) of the client's announced close time, just in case the client was about to send something as the server would have otherwise closed.
Independently, and more usefully, the server could announce its keep-alive idle time (once it exists: #14204) and the client could respect it.
The text was updated successfully, but these errors were encountered: