Everything in lws outside esp32 was changed to use lws_snprintf() a while ago.
This fixes a couple of stragglers and removes the preprocessor mangling.
In the case that the connection cannot be established because the caller
is unauthorized, it's likely they have to do something to gain
authorization before retrying. This change introduces a new message that
can be checked for to understand more about why the connection has
failed to establish.
Closes#1200
If the server rejects the attempt to establish a connection by returning
a response status other than 101, then it will not include the
Sec-WebSocket-Accept header. We need to check for 101 status (and return
an appropriate error message) before looking for the accept header.
See #1200
This allows you to set a 404 handler URL on a vhost.
The necessary user code looks like...
info.error_document_404 = "/404.html";
... at vhost-creation time.
In the existing lws_return_http_status() api, if it sees
the vhost has an "error_document_404" path set and that
we are trying to report a 404, it changes the action
instead to a redirect to the error_document_404 path.
The redirect target is returned using 404 status code.
If the redirect target doesn't exist, then it falls back
to just reporting the simple canned 404.
Re-use wsi->preamble_rx to also hold leftover rx after dealing with POST
body. Ensure ah->rx is always big enough to cope with what may have
been read into the pt->serv_buf.
Update the check for forced needed to also accept non-NULL wsi->preamble
as well as ah->rxpos != ah->rxlen as indication forced needed.
Disable autoservice on ah reset during transaction completed... it may
close the wsi underneath us when it sees and processes the pending
wsi->preamble_rx recursively otherwise.
