2004-08-17
2004-08-17 15:16:16 by Henrik Grubbström (Grubba) <grubba@grubba.org>
-
e29ec56cbff45c161aa3176b37f1382dac55a09b
(14 lines)
(+9/-5)
[
Show
| Annotate
]
Branch: 5.2
Fixed excessive "HTTP piping, but no pipe for" when running with FD_DEBUG.
Reverted misdirected kludge for Opera.
Rev: server/protocols/http.pike:1.462
2:
// Modified by Francesco Chemolli to add throttling capabilities.
// Copyright © 1996 - 2004, Roxen IS.
- constant cvs_version = "$Id: http.pike,v 1.461 2004/08/11 13:13:22 grubba Exp $";
+ constant cvs_version = "$Id: http.pike,v 1.462 2004/08/17 15:16:16 grubba Exp $";
// #define REQUEST_DEBUG
#define MAGIC_ERROR
326: Inside #if defined(FD_DEBUG)
MARK_FD("HTTP really handled, piping response");
#ifdef FD_DEBUG
call_out(timer, 30, predef::time(1)); // Update FD with time...
- #endif
+ pipe->set_done_callback(lambda (int fsent) {
+ remove_call_out(timer);
+ do_log(fsent);
+ } );
+ #else
pipe->set_done_callback( do_log );
-
+ #endif
pipe->output( my_fd );
} else {
MARK_FD("HTTP really handled, pipe done");
1844:
// Some browsers, e.g. Netscape 4.7, don't trust a zero
// content length when using keep-alive. So let's force a
// close in that case.
- // Opera 7.54/Solaris seems to have the same problem.
- // (Observed with 304's).
- if( file->error/100 >= 2 && file->len <= 0 )
+ if( file->error/100 == 2 && file->len <= 0 )
{
heads->Connection = "close";
misc->connection = "close";