-*** Version 3.0.17 Stable ***
-
-- Fixed last-chunk-detection for responses where the content was small
- enough to be read with the body, causing Privoxy to wait for the
- end of the content until the server closed the connection or the
- request timed out. Reported by "Karsten" in #3028326.
-- Responses with status code 204 weren't properly detected as body-less
- like RFC2616 mandates. Like the previous bug, this caused Privoxy to
- wait for the end of the content until the server closed the connection
- or the request timed out. Fixes #3022042 and #3025553, reported by a
- user with no visible name. Most likely also fixes a bunch of other
- AJAX-related problem reports that got closed in the past due to
- insufficient information and lack of feedback.
-- Fixed an ACL bug that made it impossible to build a blacklist.
- Usually the ACL directives are used in a whitelist, which worked
- as expected, but blacklisting is still useful for public proxies
- where one only needs to deny known abusers access.
-- Added LOG_LEVEL_RECEIVED to log the not-yet-parsed data read from the
- network. This should make debugging various parsing issues a lot easier.
-- The IPv6 code is enabled by default on Windows versions that support it.
- Patch submitted by oCameLo in #2942729.
-- In mingw32 versions, the user.filter file is reachable through the
- GUI, just like default.filter is. Feature request 3040263.
-- Added the configure option --enable-large-file-support to set a few
- defines that are required by platforms like GNU/Linux to support files
- larger then 2GB. Mainly interesting for users without proper logfile
- management.
-- Logging with "debug 16" no longer stops at the first nul byte which is
- pretty useless. Non-printable characters are replaced with their hex value
- so the result can't span multiple lines making parsing them harder then
- necessary.
-- Privoxy logs when reading an action, filter or trust file.
-- Fixed incorrect regression test markup which caused a test in
- 3.0.16 to fail while Privoxy itself was working correctly.
- While Privoxy accepts hide-referer, too, the action name is actually
- hide-referrer which is also the name used one the final results page,
- where the test expected the alias.
-
-- CGI interface improvements:
- - In finish_http_response(), continue to add the 'Connection: close'
- header if the client connection will not be kept alive.
- Anonymously pointed out in #2987454.
- - Apostrophes in block messages no longer cause parse errors
- when the blocked page is viewed with JavaScript enabled.
- Reported by dg1727 in #3062296.
- - Fix a bunch of anchors that used underscores instead of dashes.
- - Allow to keep the client connection alive after crunching the previous request.
- Already opened server connections can be kept alive, too.
- - In cgi_show_url_info(), don't forget to prefix URLs that only contain
- http:// or https:// in the path. Fixes #2975765 reported by Adam Piggott.
- - Show the 404 CGI page if cgi_send_user_manual() is called while
- local user manual delivery is disabled.
-
-- Action file improvements:
- - Enable user.filter by default. Suggested by David White in #3001830.
- - Block .sitestat.com/. Reported by johnd16 in #3002725.
- - Block .atemda.com/. Reported by johnd16 in #3002723.
- - Block js.adlink.net/. Reported by johnd16 in #3002720.
- - Block .analytics.yahoo.com/. Reported by johnd16 in #3002713.
- - Block sb.scorecardresearch.com, too. Reported by dg1727 in #2992652.
- - Fix problems noticed on Yahoo mail and news pages.
- - Remove the too broad yahoo section, only keeping the
- fast-redirects exception as discussed on ijbswa-devel@.
- - Don't block adesklets.sourceforge.net. Reported in #2974204.
- - Block chartbeat ping tracking. Reported in #2975895.
- - Tag CSS and image requests with cautious and medium settings, too.
- - Don't handle view.atdmt.com as image. It's used for click-throughs
- so users should be able to "go there anyway".
- Reported by Adam Piggott in #2975927.
- - Also let the refresh-tags filter remove invalid refresh tags where
- the 'url=' part is missing. Anonymously reported in #2986382.
- While at it, update the description to mention the fact that only
- refresh tags with refresh times above 9 seconds are covered.
- - javascript needs to be blocked with +handle-as-empty-document to
- work around Firefox bug 492459. So move .js blockers from
- +block{Might be a web-bug.} -handle-as-empty-document to
- +block{Might be a web-bug.} +handle-as-empty-document.
- - ijbswa-Feature Requests-3006719 - Block 160x578 Banners.
- - Block another omniture tracking domain.
- - Added a range-requests tagger.
- - Added two sections to get Flickr's Ajax interface working with
- default pre-settings. If you change the configuration to block
- cookies by default, you'll need additional exceptions.
- Reported by Mathias Homann in #3101419 and by Patrick on ijbswa-users@.
-
-- Documentation improvements:
- - Explicitly mention how to match all URLs.
- - Consistently recommend socks5 in the Tor FAQ entry and mention
- its advantage compared to socks4a. Reported by David in #2960129.
- - Slightly improve the explanation of why filtering may appear
- slower than it is.
- - Grammar fixes for the ACL section.
- - Fixed a link to the 'intercepting' entry and add another one.
- - Rename the 'Other' section to 'Mailing Lists' and reword it
- to make it clear that nobody is forced to use the trackers
- - Note that 'anonymously' posting on the trackers may not always
- be possible.
- - Suggest to enable debug 32768 when suspecting parsing problems.
-
-- Privoxy-Log-Parser improvements:
- - Gather statistics for ressources, methods, and HTTP versions
- used by the client.
- - Also gather statistics for blocked and redirected requests.
- - Provide the percentage of keep-alive offers the client accepted.
- - Add a --url-statistics-threshold option.
- - Add a --host-statistics-threshold option to also gather
- statistics about how many request where made per host.
- - Fix a bug in handle_loglevel_header() where a 'scan: ' got lost.
- - Add a --shorten-thread-ids option to replace the thread id with
- a decimal number.
- - Accept and ignore: Looks like we got the last chunk together
- with the server headers. We better stop reading.
- - Accept and ignore: Continue hack in da house.
- - Accept and higlight: Rejecting connection from 10.0.0.2.
- Maximum number of connections reached.
- - Accept and highlight: Loading actions file: /usr/local/etc/privoxy/default.action
- - Accept and highlight: Loading filter file: /usr/local/etc/privoxy/default.filter
- - Accept and highlight: Killed all-caps Host header line: HOST: bestproxydb.com
- - Accept and highlight: Reducing expected bytes to 0. Marking
- the server socket tainted after throwing 4 bytes away.
- - Accept: Merged multiple header lines to: 'X-FORWARDED-PROTO: http X-HOST: 127.0.0.1'
-
-- Code cleanups:
- - Remove the next member from the client_state struct. Only the main
- thread needs access to all client states so give it its own struct.
- - Garbage-collect request_contains_null_bytes().
- - Ditch redundant code in unload_configfile().
- - Ditch LogGetURLUnderCursor() which doesn't seem to be used anywhere.
- - In write_socket(), remove the write-only variable write_len in
- an ifdef __OS2__ block. Spotted by cppcheck.
- - In connect_to(), don't declare the variable 'flags' on OS/2 where
- it isn't used. Spotted by cppcheck.
- - Limit the scope of various variables. Spotted by cppcheck.
- - In add_to_iob(), turn an interestingly looking for loop into a
- boring while loop.
- - Code cleanup in preparation for external filters.
- - In listen_loop(), mention the socket on which we accepted the
- connection, not just the source IP address.
- - In write_socket(), also log the socket we're writing to.
- - In log_error(), assert that escaped characters get logged
- completely or not at all.
- - In log_error(), assert that ival and sval have reasonable values.
- There's no reason not to abort() if they don't.
- - Remove an incorrect cgi_error_unknown() call in a
- cannnot-happen-situation in send_crunch_response().
- - Clean up white-space in http_response definition and
- move the crunch_reason to the beginning.
- - Turn http_response.reason into an enum and rename it
- to http_response.crunch_reason.
- - Silence a 'gcc (Debian 4.3.2-1.1) 4.3.2' warning on i686 GNU/Linux.
- - Fix white-space in a log message in remove_chunked_transfer_coding().
- While at it, add a note that the message doesn't seem to
- be entirely correct and should be improved later on.
-
-- GNUmakefile improvements:
- - Use $(SSH) instead of ssh, so one only needs to specify a username once.
- - Removed references to the action feedback thingy that hasn't been
- working for years.
- - Consistently use shell.sourceforge.net instead of shell.sf.net so
- one doesn't need to check server fingerprints twice.
- - Removed GNUisms in the webserver and webactions targets so they
- work with standard tar.
+*** Version 3.0.20 Beta ***
+
+- Bug fixes:
+ - Client sockets are now properly shutdown and drained before being
+ closed. This fixes page truncation issues with clients that aggressively
+ pipeline data on platforms that otherwise discard already written data.
+ The issue mainly affected Opera users and was initially reported
+ by Kevin in #3464439, szotsaki provided additional information to track
+ down the cause.
+ - Fix latency calculation for shared connections (disabled by default).
+ It was broken since their introduction in 2009. The calculated latency
+ for most connections would be 0 in which case the timeout detection
+ failed to account for the real latency.
+ - Reject URLs with invalid port. Previously they were parsed incorrectly and
+ characters between the port number and the first slash were silently
+ dropped as shown by curl test 187.
+ - The default-server-timeout and socket-timeout directives accept 0 as
+ valid value.
+ - Fix a race condition on Windows that could cause Privoxy to become
+ unresponsive after toggling it on or off through the taskbar icon.
+ Reported by Tim H. in #3525694.
+ - Fix the compilation on Windows when configured without IPv6 support.
+ - Fix an assertion that could cause debug builds to abort() in case of
+ socks5 connection failures with "debug 2" enabled.
+ - Fix an assertion that could cause debug builds to abort() if a filter
+ contained nul bytes in the replacement text.
+
+- General improvements:
+ - Significantly improved keep-alive support for both client and server
+ connections.
+ - New debug log level 65536 which logs all actions that were applied to
+ the request.
+ - New directive client-header-order to forward client headers in a
+ different order than the one in which they arrived.
+ - New directive tolerate-pipelining to allow client-side pipelining.
+ If enabled (3.0.20 beta enables it by default), Privoxy will keep
+ pipelined client requests around to deal with them once the current
+ request has been served.
+ - New --config-test option to let Privoxy exit after checking whether or not
+ the configuration seems valid. The limitations noted in TODO #22 and #23
+ still apply. Based on a patch by Ramkumar Chinchani.
+ - New limit-cookie-lifetime{} action to let cookies expire before the end
+ of the session. Suggested by Rick Sykes in #1049575.
+ - Increase the hard-coded maximum number of actions and filter files from
+ 10 to 30 (each). It doesn't significantly affect Privoxy's memory usage
+ and recompiling wasn't an option for all Privoxy users that reached the
+ limit.
+ - Add support for chunk-encoded client request bodies. Previously
+ chunk-encoded request bodies weren't guaranteed to be forwarded correctly,
+ so this can also be considered a bug fix although chunk-encoded request
+ bodies aren't commonly used in the real world.
+ - Add support for Tor's optimistic-data SOCKS extension, which can reduce the
+ latency for requests on newly created connections. Currently only the
+ headers are sent optimistically and only if the client request has already
+ been read completely which rules out requests with large bodies.
+ - After preventing the client from pipelining, don't signal keep-alive
+ intentions. When looking at the response headers alone, it previously
+ wasn't obvious from the client's perspective that no additional responses
+ should be expected.
+ - Stop considering client sockets tainted after receving a request with body.
+ It hasn't been necessary for a while now and unnecessarily causes test
+ failures when using curl's test suite.
+ - Allow HTTP/1.0 clients to signal interest in keep-alive through the
+ Proxy-Connection header. While such client are rare in the real world, it
+ doesn't hurt and couple of curl tests rely on it.
+ - Only remove duplicated Content-Type headers when filters are enabled.
+ If they are not it doesn't cause ill effects and the user might not want it.
+ Downgrade the removal message to LOG_LEVEL_HEADER to clarify that it's not
+ an error in Privoxy and is unlikely to cause any problems in general.
+ Anonymously reported in #3599335.
+ - Set the socket option SO_LINGER for the client socket.
+ - Move several variable declarations to the beginning of their code block.
+ It's required when compiling with gcc 2.95 which is still used on some
+ platforms. Initial patch submitted by Simon South in #3564815.
+ - Optionally try to sanity-check strptime() results before trusting them.
+ Broken strptime() implementations have caused problems in the past and
+ the most recent offender seems to be FreeBSD's libc (standards/173421).
+ - When filtering is enabled, let Range headers pass if the range starts at
+ the beginning. This should work around (or at least reduce ) the video
+ playback issues with various Apple clients as reported by Duc in #3426305.
+ - Do not confuse a client hanging up with a connection time out. If a client
+ closes its side of the connection without sending a request line, do not
+ send the CLIENT_CONNECTION_TIMEOUT_RESPONSE, but report the condition
+ properly.
+ - Allow closing curly braces as part of action values as long as they are
+ escaped.
+ - On Windows, the logfile is now written before showing the GUI error
+ message which blocks until the user acknowledges it.
+ Reported by Adriaan in #3593603.
+ - Remove an unreasonable parameter limit in the CGI interface. The new
+ parameter limit depends on the memory available and is currently unlikely
+ to be reachable, due to other limits in both Privoxy and common clients.
+ Reported by Andrew on ijbswa-users@.
+ - Decrease the chances of parse failures after requests with unsupported
+ methods were sent to the CGI interface.
+
+*** Version 3.0.19 Stable ***
+
+- Bug fixes:
+ - Prevent a segmentation fault when de-chunking buffered content.
+ It could be triggered by malicious web servers if Privoxy was
+ configured to filter the content and running on a platform
+ where SIZE_T_MAX isn't larger than UINT_MAX, which probably
+ includes most 32-bit systems. On those platforms, all Privoxy
+ versions before 3.0.19 appear to be affected.
+ To be on the safe side, this bug should be presumed to allow
+ code execution as proving that it doesn't seems unrealistic.
+ - Do not expect a response from the SOCKS4/4A server until it
+ got something to respond to. This regression was introduced
+ in 3.0.18 and prevented the SOCKS4/4A negotiation from working.
+ Reported by qqqqqw in #3459781.
+
+- General improvements:
+ - Fix an off-by-one in an error message about connect failures.
+ - Use a GNUMakefile variable for the webserver root directory and
+ update the path. Sourceforge changed it which broke various
+ web-related targets.
+ - Update the CODE_STATUS description.