privoxy.git
12 years agoEnable FEATURE_CONNECTION_KEEP_ALIVE unconditionally. Enable FEATURE_CONNECTION_SHARI...
Fabian Keil [Sun, 6 Sep 2009 14:17:00 +0000 (14:17 +0000)]
Enable FEATURE_CONNECTION_KEEP_ALIVE unconditionally. Enable FEATURE_CONNECTION_SHARING on platforms with supported mutexes.

12 years agoDisable connection-sharing unless FEATURE_CONNECTION_SHARING is defined.
Fabian Keil [Sun, 6 Sep 2009 14:15:46 +0000 (14:15 +0000)]
Disable connection-sharing unless FEATURE_CONNECTION_SHARING is defined.

12 years agoIn get_content_length(), add a sanity check for mingw32.
Fabian Keil [Sun, 6 Sep 2009 14:11:06 +0000 (14:11 +0000)]
In get_content_length(), add a sanity check for mingw32.

12 years agoFactor get_content_length() out of client_save_content_length() and server_save_conte...
Fabian Keil [Sun, 6 Sep 2009 14:10:07 +0000 (14:10 +0000)]
Factor get_content_length() out of client_save_content_length() and server_save_content_length().

12 years agoIn get_reusable_connection(), also log when the timestamp was made, the timeout and...
Fabian Keil [Sun, 6 Sep 2009 14:09:19 +0000 (14:09 +0000)]
In get_reusable_connection(), also log when the timestamp was made, the timeout and the assumed latency.

12 years agoIn serve(), don't bother assuming a timeout value if we aren't going to reuse the...
Fabian Keil [Sun, 6 Sep 2009 14:07:56 +0000 (14:07 +0000)]
In serve(), don't bother assuming a timeout value if we aren't going to reuse the connection anyway.

While it didn't hurt, the log message looked somewhat stupid.

12 years agoSimplify the initial byte_count calculation to get rid of the lame workaround added...
Fabian Keil [Sat, 5 Sep 2009 18:04:37 +0000 (18:04 +0000)]
Simplify the initial byte_count calculation to get rid of the lame workaround added in the previous commit.

12 years agoIn chat(), reset byte_count each time we are buffering additional server headers.
Fabian Keil [Fri, 4 Sep 2009 18:28:32 +0000 (18:28 +0000)]
In chat(), reset byte_count each time we are buffering additional server headers.

We otherwise seem to miscalculate it, if we don't get the whole head in two reads or less.
This is another attempt to fix #2840156 reported by Olivier.
It's not a permanent solution but should do for now.

12 years agoDon't miscalculate byte_count when buffering server headers without any complete...
Fabian Keil [Fri, 28 Aug 2009 15:45:18 +0000 (15:45 +0000)]
Don't miscalculate byte_count when buffering server headers without any complete server header read yet.

Likely to fix #2840156 reported by Oliver.

12 years agoFactor verify_request_length() out of parse_client_request().
Fabian Keil [Fri, 28 Aug 2009 14:42:06 +0000 (14:42 +0000)]
Factor verify_request_length() out of parse_client_request().

While at it, check the request length for all but CONNECT requests,
independently of CSP_FLAG_CLIENT_CONNECTION_KEEP_ALIVE.

12 years agoAccept and highlight: Reduced expected bytes to 0 to account for the 1542 ones we...
Fabian Keil [Thu, 20 Aug 2009 15:43:56 +0000 (15:43 +0000)]
Accept and highlight: Reduced expected bytes to 0 to account for the 1542 ones we already got.

12 years agoRemove redundant log message in client_save_content_length().
Fabian Keil [Thu, 20 Aug 2009 15:27:03 +0000 (15:27 +0000)]
Remove redundant log message in client_save_content_length().

12 years agoHighlight: Empty server or forwarder response received on socket 4.
Fabian Keil [Thu, 20 Aug 2009 14:19:52 +0000 (14:19 +0000)]
Highlight: Empty server or forwarder response received on socket 4.

12 years agoAccept and ignore: There might be a request body. The connection will not be kept...
Fabian Keil [Thu, 20 Aug 2009 14:18:27 +0000 (14:18 +0000)]
Accept and ignore: There might be a request body. The connection will not be kept alive.

12 years agoIn chat(), set CSP_FLAG_CLIENT_REQUEST_COMPLETELY_READ after csp->expected_client_con...
Fabian Keil [Wed, 19 Aug 2009 16:02:53 +0000 (16:02 +0000)]
In chat(), set CSP_FLAG_CLIENT_REQUEST_COMPLETELY_READ after csp->expected_client_content_length reaches zero so we don't have to deal with the next request until we're ready again.

12 years agoSubtract already buffered request bytes from the expected ones and throw away overrea...
Fabian Keil [Wed, 19 Aug 2009 16:00:07 +0000 (16:00 +0000)]
Subtract already buffered request bytes from the expected ones and throw away overread bytes in pipelined requests with body to let the client retry on an untainted socket.

12 years agoIf the client specified a content length, assume that there will be a body and don...
Fabian Keil [Wed, 19 Aug 2009 15:59:02 +0000 (15:59 +0000)]
If the client specified a content length, assume that there will be a body and don't bother reading more than the client announced.

12 years agoIn preparation to allow connection reuse after forwarding POST requests, save the...
Fabian Keil [Wed, 19 Aug 2009 15:57:13 +0000 (15:57 +0000)]
In preparation to allow connection reuse after forwarding POST requests, save the value of the Content-Length header sent by the client.

12 years agoFix gcc45 warning in pcrs_substitute().
Fabian Keil [Wed, 19 Aug 2009 15:28:08 +0000 (15:28 +0000)]
Fix gcc45 warning in pcrs_substitute().

12 years agoRemove bogus assertion in decompress_iob().
Fabian Keil [Wed, 19 Aug 2009 15:26:36 +0000 (15:26 +0000)]
Remove bogus assertion in decompress_iob().

Unsigned variables rarely become negative.

12 years agoFix gcc45 warnings in decompress_iob().
Fabian Keil [Wed, 19 Aug 2009 15:25:31 +0000 (15:25 +0000)]
Fix gcc45 warnings in decompress_iob().

12 years agoFix log message in case of DNS resolution problems in acl_addr().
Fabian Keil [Wed, 19 Aug 2009 15:24:30 +0000 (15:24 +0000)]
Fix log message in case of DNS resolution problems in acl_addr().

12 years agoFix gcc45 warning.
Fabian Keil [Wed, 19 Aug 2009 15:22:18 +0000 (15:22 +0000)]
Fix gcc45 warning.

12 years agoAdd redirect exceptions for some news sites linked from www.ni.dk.
Fabian Keil [Sat, 15 Aug 2009 16:24:39 +0000 (16:24 +0000)]
Add redirect exceptions for some news sites linked from ni.dk.

Anonymously reported in #2830392.

12 years agoMake a few domain patterns a bit more restrictive.
Fabian Keil [Sat, 15 Aug 2009 16:12:16 +0000 (16:12 +0000)]
Make a few domain patterns a bit more restrictive.

12 years agoUpdate the "Download recent releases" URL and fix the "Pictures of the Privoxy Team...
Fabian Keil [Sat, 1 Aug 2009 11:48:22 +0000 (11:48 +0000)]
Update the "Download recent releases" URL and fix the "Pictures of the Privoxy Team" URL.
privoxy.org redirects to www.privoxy.org anyway.

12 years agoFix server_set_cookie() description. The jarfile code is gone.
Fabian Keil [Sat, 1 Aug 2009 11:46:59 +0000 (11:46 +0000)]
Fix server_set_cookie() description. The jarfile code is gone.

12 years agoFix comment.
Fabian Keil [Sat, 1 Aug 2009 11:42:43 +0000 (11:42 +0000)]
Fix comment.

12 years agoBump version to 3.0.15 UNRELEASED.
Fabian Keil [Sat, 1 Aug 2009 11:41:58 +0000 (11:41 +0000)]
Bump version to 3.0.15 UNRELEASED.

12 years agoThis announcement is for 3.0.14 beta, not 3.0.13 beta.
Fabian Keil [Fri, 24 Jul 2009 14:15:10 +0000 (14:15 +0000)]
This announcement is for 3.0.14 beta, not 3.0.13 beta.

12 years agoRebuild for Privoxy 3.0.14 beta.
Fabian Keil [Fri, 24 Jul 2009 12:22:59 +0000 (12:22 +0000)]
Rebuild for Privoxy 3.0.14 beta.

12 years agoRemove duplicated period.
Fabian Keil [Fri, 24 Jul 2009 12:20:30 +0000 (12:20 +0000)]
Remove duplicated period.

12 years agoChange code status from BETA to beta.
Fabian Keil [Fri, 24 Jul 2009 11:48:08 +0000 (11:48 +0000)]
Change code status from BETA to beta.

12 years agoBump version for 3.0.14 beta.
Fabian Keil [Thu, 23 Jul 2009 15:16:16 +0000 (15:16 +0000)]
Bump version for 3.0.14 beta.

12 years agoDisarm the !socket_is_still_usable() check in chat() for mingw32 where it seems to...
Fabian Keil [Wed, 22 Jul 2009 22:31:54 +0000 (22:31 +0000)]
Disarm the !socket_is_still_usable() check in chat() for mingw32 where it seems to misfire.

Pointed out by Lee.

12 years agoLet socket_is_still_usable() properly report the socket status if there's data waiting.
Fabian Keil [Wed, 22 Jul 2009 22:27:16 +0000 (22:27 +0000)]
Let socket_is_still_usable() properly report the socket status if there's data waiting.

Pointed out by Lee.

12 years agoalso hilight
Lee [Tue, 21 Jul 2009 16:29:57 +0000 (16:29 +0000)]
also hilight
  Found reusable socket
  Created new connection
messages in the log console window.

12 years agoDisable content filtering for URLs ending in .ogg or ogv.
Fabian Keil [Tue, 21 Jul 2009 14:19:57 +0000 (14:19 +0000)]
Disable content filtering for URLs ending in .ogg or ogv.
Fixes #2816708 reported by cyberpatrol and Johan Brannlund.

12 years agoIgnore line feeds at the end of the message.
Fabian Keil [Tue, 21 Jul 2009 14:04:25 +0000 (14:04 +0000)]
Ignore line feeds at the end of the message.

12 years agoHighlight: Connect: to archiv.radiotux.de
Fabian Keil [Tue, 21 Jul 2009 13:55:59 +0000 (13:55 +0000)]
Highlight: Connect: to archiv.radiotux.de

12 years agoIf RUNTIME_FEATURE_CONNECTION_KEEP_ALIVE is disabled, don't let 'Connection: keep...
Fabian Keil [Sun, 19 Jul 2009 11:48:32 +0000 (11:48 +0000)]
If RUNTIME_FEATURE_CONNECTION_KEEP_ALIVE is disabled, don't let 'Connection: keep-alive' headers through.

12 years agoIf you're masking csp->flags with RUNTIME_FEATURE_CONNECTION_KEEP_ALIVE you're doing...
Fabian Keil [Sun, 19 Jul 2009 11:19:50 +0000 (11:19 +0000)]
If you're masking csp->flags with RUNTIME_FEATURE_CONNECTION_KEEP_ALIVE you're doing it wrong.

12 years agoIf connection-sharing is enabled but keep-alive-timeout isn't, log a warning and...
Fabian Keil [Sun, 19 Jul 2009 10:07:46 +0000 (10:07 +0000)]
If connection-sharing is enabled but keep-alive-timeout isn't, log a warning and disable it.

12 years agoDon't set CSP_FLAG_CLIENT_CONNECTION_KEEP_ALIVE in client_connection() if the client...
Fabian Keil [Sun, 19 Jul 2009 10:06:33 +0000 (10:06 +0000)]
Don't set CSP_FLAG_CLIENT_CONNECTION_KEEP_ALIVE in client_connection() if the client didn't request it.

12 years agoDon't assume persistence if RUNTIME_FEATURE_CONNECTION_KEEP_ALIVE is disabled.
Fabian Keil [Sun, 19 Jul 2009 10:04:55 +0000 (10:04 +0000)]
Don't assume persistence if RUNTIME_FEATURE_CONNECTION_KEEP_ALIVE is disabled.

12 years agoUnbreak client_connection() for !defined(FEATURE_CONNECTION_KEEP_ALIVE).
Fabian Keil [Sun, 19 Jul 2009 09:24:17 +0000 (09:24 +0000)]
Unbreak client_connection() for !defined(FEATURE_CONNECTION_KEEP_ALIVE).

Won't somebody think of the OS/2 users?

12 years agoUpdate FEATURE_CONNECTION_KEEP_ALIVE description.
Fabian Keil [Sun, 19 Jul 2009 09:22:02 +0000 (09:22 +0000)]
Update FEATURE_CONNECTION_KEEP_ALIVE description.

12 years agoTests with level 16 also depends on FEATURE_CONNECTION_KEEP_ALIVE.
Fabian Keil [Sun, 19 Jul 2009 09:21:19 +0000 (09:21 +0000)]
Tests with level 16 also depends on FEATURE_CONNECTION_KEEP_ALIVE.

12 years agoThere currently are no packages for AmigaOS, Fedora or Mac OSX.
Fabian Keil [Sun, 19 Jul 2009 08:56:03 +0000 (08:56 +0000)]
There currently are no packages for AmigaOS, Fedora or Mac OSX.

12 years ago- Add ChangeLog for 3.0.14 beta.
Fabian Keil [Sat, 18 Jul 2009 18:21:13 +0000 (18:21 +0000)]
- Add ChangeLog for 3.0.14 beta.
- Sync feature list with newfeatures.sgml.

12 years agoDon't claim that NTLM should work when there are multiple reports that it doesn't.
Fabian Keil [Sat, 18 Jul 2009 18:11:12 +0000 (18:11 +0000)]
Don't claim that NTLM should work when there are multiple reports that it doesn't.

12 years agoRegenerate for 3.0.14 beta.
Fabian Keil [Sat, 18 Jul 2009 16:28:23 +0000 (16:28 +0000)]
Regenerate for 3.0.14 beta.

12 years agoFix trailing whitespace.
Fabian Keil [Sat, 18 Jul 2009 16:25:36 +0000 (16:25 +0000)]
Fix trailing whitespace.

12 years agoBump entities for 3.0.14 beta.
Fabian Keil [Sat, 18 Jul 2009 16:24:39 +0000 (16:24 +0000)]
Bump entities for 3.0.14 beta.

12 years agoUpdate for 3.0.14 beta.
Fabian Keil [Sat, 18 Jul 2009 16:06:33 +0000 (16:06 +0000)]
Update for 3.0.14 beta.

12 years agoUpdate description for 3.0.14 beta.
Fabian Keil [Sat, 18 Jul 2009 16:00:17 +0000 (16:00 +0000)]
Update description for 3.0.14 beta.

12 years agoAdd most of the changes in 3.0.14 to the "What's New" section.
Fabian Keil [Sat, 18 Jul 2009 15:49:23 +0000 (15:49 +0000)]
Add most of the changes in 3.0.14 to the "What's New" section.

12 years ago- Move less interesting changes to the bottom.
Fabian Keil [Sat, 18 Jul 2009 14:59:25 +0000 (14:59 +0000)]
- Move less interesting changes to the bottom.
- Cosmetics.

12 years agoAdd changes in 3.0.14 BETA.
Fabian Keil [Sat, 18 Jul 2009 12:22:44 +0000 (12:22 +0000)]
Add changes in 3.0.14 BETA.

12 years agoChange code status to BETA.
Fabian Keil [Sat, 18 Jul 2009 12:21:03 +0000 (12:21 +0000)]
Change code status to BETA.

12 years agoDon't add a Proxy-Connection header if there already is one.
Fabian Keil [Sat, 18 Jul 2009 12:20:05 +0000 (12:20 +0000)]
Don't add a Proxy-Connection header if there already is one.

12 years agoDon't describe the action files in the FAQ when we can simply link to the description...
Fabian Keil [Sat, 18 Jul 2009 12:18:52 +0000 (12:18 +0000)]
Don't describe the action files in the FAQ when we can simply link to the description in the User Manual.

12 years agoAccept and ignore: The server didn't specify how long the connection will stay open...
Fabian Keil [Tue, 14 Jul 2009 18:40:48 +0000 (18:40 +0000)]
Accept and ignore: The server didn't specify how long the connection will stay open. Assume it's only a second.

12 years agoIn parse_client_request(), assume a request is complete if we got the whole header...
Fabian Keil [Tue, 14 Jul 2009 18:02:25 +0000 (18:02 +0000)]
In parse_client_request(), assume a request is complete if we got the whole header and the method doesn't imply a body.

Should break less non-traditional requests than the previous hack.

12 years agoIf the server doesn't specify how long the connection stays alive, err on the safe...
Fabian Keil [Tue, 14 Jul 2009 17:50:34 +0000 (17:50 +0000)]
If the server doesn't specify how long the connection stays alive, err on the safe side of caution and assume it's only a second.

12 years agoBack out the last commit so it can return with related changes.
Fabian Keil [Tue, 14 Jul 2009 17:45:05 +0000 (17:45 +0000)]
Back out the last commit so it can return with related changes.
The bug it fixes hasn't actually been committed.

12 years agoMust not reuse flag masks.
Fabian Keil [Mon, 13 Jul 2009 19:16:15 +0000 (19:16 +0000)]
Must not reuse flag masks.

12 years agoAccept and ignore: Removing 'Connection: close' to imply keep-alive.
Fabian Keil [Mon, 13 Jul 2009 17:17:47 +0000 (17:17 +0000)]
Accept and ignore: Removing 'Connection: close' to imply keep-alive.

12 years agoTake the latency into account when evaluating whether or not to reuse a connection.
Fabian Keil [Mon, 13 Jul 2009 17:12:28 +0000 (17:12 +0000)]
Take the latency into account when evaluating whether or not to reuse a connection.

12 years agoIf keep-alive support is disabled, don't let the client's Keep-Alive header through.
Fabian Keil [Mon, 13 Jul 2009 17:10:57 +0000 (17:10 +0000)]
If keep-alive support is disabled, don't let the client's Keep-Alive header through.

Anonymously reported in #2818986.

12 years agoTimestamp the connection at the end of chat().
Fabian Keil [Mon, 13 Jul 2009 17:08:41 +0000 (17:08 +0000)]
Timestamp the connection at the end of chat().

The timeout doesn't start until the whole request has been received.

12 years agoIf the socket isn't reusable, don't bother remembering the connection.
Fabian Keil [Mon, 13 Jul 2009 17:05:36 +0000 (17:05 +0000)]
If the socket isn't reusable, don't bother remembering the connection.

12 years agoAccept and ignore: The server still wants to talk, but the client hung up on us.
Fabian Keil [Sun, 12 Jul 2009 09:20:09 +0000 (09:20 +0000)]
Accept and ignore: The server still wants to talk, but the client hung up on us.

12 years agoIn the year 2525 ... header time randomization fails on mingw32.
Fabian Keil [Sat, 11 Jul 2009 16:30:39 +0000 (16:30 +0000)]
In the year 2525 ... header time randomization fails on mingw32.

12 years agoSet the timestamp directly after receiving the response headers.
Fabian Keil [Sat, 11 Jul 2009 14:49:09 +0000 (14:49 +0000)]
Set the timestamp directly after receiving the response headers.

When checking if the server socket is still usable, we have to take the
time into account it took the client to decide that it's done talking.

12 years agoDon't bother reading from the server if the client already hung up on us.
Fabian Keil [Sat, 11 Jul 2009 14:39:34 +0000 (14:39 +0000)]
Don't bother reading from the server if the client already hung up on us.

12 years agoUpdate regression tests for the keep-alive changes.
Fabian Keil [Sat, 11 Jul 2009 11:20:53 +0000 (11:20 +0000)]
Update regression tests for the keep-alive changes.

12 years agoBack out 1.187 for now. There has to be a better way.
Fabian Keil [Sat, 11 Jul 2009 11:20:12 +0000 (11:20 +0000)]
Back out 1.187 for now. There has to be a better way.

12 years agoFor HTTP/1.1 requests other than CONNECT requests, assume persistence unless the...
Fabian Keil [Sat, 11 Jul 2009 11:17:35 +0000 (11:17 +0000)]
For HTTP/1.1 requests other than CONNECT requests, assume persistence unless the client says otherwise.

12 years agoGarbage collect get_appropiate_connection_header() as we no longer use it.
Fabian Keil [Sat, 11 Jul 2009 11:16:19 +0000 (11:16 +0000)]
Garbage collect get_appropiate_connection_header() as we no longer use it.

12 years agoDon't set 'Connection: keep-alive' headers in server_connection_adder() either.
Fabian Keil [Sat, 11 Jul 2009 11:15:53 +0000 (11:15 +0000)]
Don't set 'Connection: keep-alive' headers in server_connection_adder() either.

12 years agoIn client_connection_header_adder(), add a 'Connection: close' header or none at...
Fabian Keil [Sat, 11 Jul 2009 11:15:14 +0000 (11:15 +0000)]
In client_connection_header_adder(), add a 'Connection: close' header or none at all.

Suggested by Lee.

12 years agoHashes in git object URLs may contain 'ad's. Deal with it.
Fabian Keil [Sat, 11 Jul 2009 11:13:26 +0000 (11:13 +0000)]
Hashes in git object URLs may contain 'ad's. Deal with it.

12 years agodon't need the extra time() call
Lee [Wed, 8 Jul 2009 23:18:05 +0000 (23:18 +0000)]
don't need the extra time() call

12 years agoHighlight: Client keep-alive timeout is 20. Sticking with 10.
Fabian Keil [Wed, 8 Jul 2009 17:26:46 +0000 (17:26 +0000)]
Highlight: Client keep-alive timeout is 20. Sticking with 10.

12 years agoFix a log message in client_keep_alive().
Fabian Keil [Wed, 8 Jul 2009 17:25:37 +0000 (17:25 +0000)]
Fix a log message in client_keep_alive().

12 years agoRevert r261 as it would sabotage NTLM authentication.
Fabian Keil [Tue, 7 Jul 2009 16:42:26 +0000 (16:42 +0000)]
Revert r261 as it would sabotage NTLM authentication.

12 years agoVerizon DNS server returns addr of their 'search' page instead of domain not found.
Lee [Tue, 7 Jul 2009 16:35:01 +0000 (16:35 +0000)]
Verizon DNS server returns addr of their 'search' page instead of domain not found.

12 years agoFixes Bugs item #2816541 - http download stops at 0.3GB (mingw32)
Lee [Sun, 5 Jul 2009 20:43:14 +0000 (20:43 +0000)]
Fixes Bugs item #2816541 - http download stops at 0.3GB (mingw32)
  Microsoft has their own way - %I64U instead of %llu for long long unsigned.

12 years agoMention the tainted socket in mark_server_socket_tainted().
Fabian Keil [Sun, 5 Jul 2009 13:24:04 +0000 (13:24 +0000)]
Mention the tainted socket in mark_server_socket_tainted().

12 years agoHighlight: Marking the server socket 7 tainted.
Fabian Keil [Sun, 5 Jul 2009 12:11:18 +0000 (12:11 +0000)]
Highlight: Marking the server socket 7 tainted.

12 years agoAlso react to the client's Keep-Alive header.
Fabian Keil [Sun, 5 Jul 2009 12:04:46 +0000 (12:04 +0000)]
Also react to the client's Keep-Alive header.

12 years agoComment spelling fix.
Fabian Keil [Sun, 5 Jul 2009 12:02:54 +0000 (12:02 +0000)]
Comment spelling fix.

12 years agoMention the offending socket in the 'Empty server or forwarder response' message.
Fabian Keil [Sun, 5 Jul 2009 12:02:25 +0000 (12:02 +0000)]
Mention the offending socket in the 'Empty server or forwarder response' message.

12 years agoIf we decided not to keep the connection to the server alive, don't reuse a remembere...
Fabian Keil [Sun, 5 Jul 2009 12:01:45 +0000 (12:01 +0000)]
If we decided not to keep the connection to the server alive, don't reuse a remembered connection either.

12 years agoIf we decided not to keep the connection to the server alive, but have a reusable...
Fabian Keil [Sun, 5 Jul 2009 12:00:53 +0000 (12:00 +0000)]
If we decided not to keep the connection to the server alive, but have a reusable connection open, close it and create a new one.

12 years agoIn mark_server_socket_tainted(), update the log message and stop unsetting CSP_FLAG_S...
Fabian Keil [Sun, 5 Jul 2009 12:00:09 +0000 (12:00 +0000)]
In mark_server_socket_tainted(), update the log message and stop unsetting CSP_FLAG_SERVER_CONNECTION_KEEP_ALIVE which is no longer necessary.

13 years agoAccept the recently added LOG_LEVEL_CONNECT messages:
Fabian Keil [Tue, 30 Jun 2009 19:08:37 +0000 (19:08 +0000)]
Accept the recently added LOG_LEVEL_CONNECT messages:
Complete client request received.
Possible pipeline attempt detected. The connection will not be kept alive and we will only serve the first request.
POST request detected. The connection will not be kept alive.

13 years agoIf the status code isn't 200, don't assume a missing Connection header for a HTTP...
Fabian Keil [Tue, 30 Jun 2009 18:32:04 +0000 (18:32 +0000)]
If the status code isn't 200, don't assume a missing Connection header for a HTTP/1.1 response implies keep-alive.

Seems to work around problems with some servers.

13 years agoIf we think we already got a complete request, don't try to read from the client...
Fabian Keil [Sun, 28 Jun 2009 14:31:43 +0000 (14:31 +0000)]
If we think we already got a complete request, don't try to read from the client until we're ready again.

If we notice that we got more than a single request, mark
the server socket tainted and only serve the first request.