X-Git-Url: http://www.privoxy.org/gitweb/?a=blobdiff_plain;f=doc%2Fwebserver%2Fuser-manual%2Fwhatsnew.html;h=3dfd28640bc6e11da81804c4e59f92e5e8c769c2;hb=afdf7f7b2cbdff9a333ac73e0bdb74474015705d;hp=d9c7b2d2aef6135350352d24145f02ff1dbb0239;hpb=12b554fe26edd92957aaaec82f7505cb3d4bb985;p=privoxy.git

diff --git a/doc/webserver/user-manual/whatsnew.html b/doc/webserver/user-manual/whatsnew.html
index d9c7b2d2..3dfd2864 100644
--- a/doc/webserver/user-manual/whatsnew.html
+++ b/doc/webserver/user-manual/whatsnew.html
@@ -6,7 +6,7 @@
   <title>What's New in this Release</title>
   <meta name="GENERATOR" content=
   "Modular DocBook HTML Stylesheet Version 1.79">
-  <link rel="HOME" title="Privoxy 3.0.20 User Manual" href="index.html">
+  <link rel="HOME" title="Privoxy 3.0.24 User Manual" href="index.html">
   <link rel="PREVIOUS" title="Installation" href="installation.html">
   <link rel="NEXT" title="Quickstart to Using Privoxy" href=
   "quickstart.html">
@@ -21,7 +21,7 @@
     <table summary="Header navigation table" width="100%" border="0"
     cellpadding="0" cellspacing="0">
       <tr>
-        <th colspan="3" align="center">Privoxy 3.0.20 User Manual</th>
+        <th colspan="3" align="center">Privoxy 3.0.24 User Manual</th>
       </tr>
 
       <tr>
@@ -41,869 +41,271 @@
     <h1 class="SECT1"><a name="WHATSNEW" id="WHATSNEW">3. What's New in this
     Release</a></h1>
 
-    <p><span class="APPLICATION">Privoxy 3.0.20</span> is a beta release. The
-    changes since 3.0.19 stable are:</p>
+    <p><span class="APPLICATION">Privoxy 3.0.24</span> stable contains a
+    couple of new features but is mainly a bug-fix release. Two of the fixed
+    bugs are security issues and may be used to remotely trigger crashes on
+    platforms that carefully check memory accesses (most don't).</p>
 
     <ul>
       <li>
-        <p>Bug fixes:</p>
+        <p>Security fixes (denial of service):</p>
 
         <ul>
           <li>
-            <p>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.</p>
-          </li>
-
-          <li>
-            <p>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.</p>
-          </li>
-
-          <li>
-            <p>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.</p>
-          </li>
-
-          <li>
-            <p>The default-server-timeout and socket-timeout directives
-            accept 0 as valid value.</p>
-          </li>
-
-          <li>
-            <p>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.</p>
-          </li>
-
-          <li>
-            <p>Fix the compilation on Windows when configured without IPv6
-            support.</p>
-          </li>
-
-          <li>
-            <p>Fix an assertion that could cause debug builds to abort() in
-            case of socks5 connection failures with "debug 2" enabled.</p>
+            <p>Prevent invalid reads in case of corrupt chunk-encoded
+            content. CVE-2016-1982. Bug discovered with afl-fuzz and
+            AddressSanitizer.</p>
           </li>
 
           <li>
-            <p>Fix an assertion that could cause debug builds to abort() if a
-            filter contained nul bytes in the replacement text.</p>
+            <p>Remove empty Host headers in client requests. Previously they
+            would result in invalid reads. CVE-2016-1983. Bug discovered with
+            afl-fuzz and AddressSanitizer.</p>
           </li>
         </ul>
       </li>
 
       <li>
-        <p>General improvements:</p>
+        <p>Bug fixes:</p>
 
         <ul>
           <li>
-            <p>Significantly improved keep-alive support for both client and
-            server connections.</p>
-          </li>
-
-          <li>
-            <p>New debug log level 65536 which logs all actions that were
-            applied to the request.</p>
-          </li>
-
-          <li>
-            <p>New directive client-header-order to forward client headers in
-            a different order than the one in which they arrived.</p>
-          </li>
-
-          <li>
-            <p>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.</p>
-          </li>
-
-          <li>
-            <p>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.</p>
-          </li>
-
-          <li>
-            <p>New limit-cookie-lifetime{} action to let cookies expire
-            before the end of the session. Suggested by Rick Sykes in
-            #1049575.</p>
-          </li>
-
-          <li>
-            <p>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.</p>
-          </li>
-
-          <li>
-            <p>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.</p>
-          </li>
-
-          <li>
-            <p>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.</p>
-          </li>
-
-          <li>
-            <p>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.</p>
-          </li>
-
-          <li>
-            <p>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.</p>
-          </li>
-
-          <li>
-            <p>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.</p>
-          </li>
-
-          <li>
-            <p>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.</p>
-          </li>
-
-          <li>
-            <p>Set the socket option SO_LINGER for the client socket.</p>
-          </li>
-
-          <li>
-            <p>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.</p>
-          </li>
-
-          <li>
-            <p>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).</p>
-          </li>
-
-          <li>
-            <p>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.</p>
+            <p>When using socks5t, send the request body optimistically as
+            well. Previously the request body wasn't guaranteed to be sent at
+            all and the error message incorrectly blamed the server. Fixes
+            #1686 reported by Peter M&uuml;ller and G4JC.</p>
           </li>
 
           <li>
-            <p>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.</p>
+            <p>Fixed buffer scaling in execute_external_filter() that could
+            lead to crashes. Submitted by Yang Xia in #892.</p>
           </li>
 
           <li>
-            <p>Allow closing curly braces as part of action values as long as
-            they are escaped.</p>
-          </li>
-
-          <li>
-            <p>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.</p>
-          </li>
-
-          <li>
-            <p>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
+            <p>Fixed crashes when executing external filters on platforms
+            like Mac OS X. Reported by Jonathan McKenzie on
             ijbswa-users@.</p>
           </li>
 
           <li>
-            <p>Decrease the chances of parse failures after requests with
-            unsupported methods were sent to the CGI interface.</p>
-          </li>
-        </ul>
-      </li>
-
-      <li>
-        <p>Action file improvements:</p>
-
-        <ul>
-          <li>
-            <p>Remove the comment that indicated that updated default.action
-            versions are released on their own.</p>
-          </li>
-
-          <li>
-            <p>Block 'optimize.indieclick.com/' and
-            'optimized-by.rubiconproject.com/'</p>
-          </li>
-
-          <li>
-            <p>Unblock 'adjamblog.wordpress.com/' and
-            'adjamblog.files.wordpress.com/'. Reported by Ryan Farmer in
-            #3496116.</p>
-          </li>
-
-          <li>
-            <p>Unblock '/.*Bugtracker'. Reported by pwhk in #3522341.</p>
-          </li>
-
-          <li>
-            <p>Add test URLs for '.freebsd.org' and '.watson.org'.</p>
-          </li>
-
-          <li>
-            <p>Unblock '.urbandictionary.com/popular'.</p>
-          </li>
-
-          <li>
-            <p>Block '.adnxs.com/'.</p>
-          </li>
-
-          <li>
-            <p>Block 'farm.plista.com/widgetdata.php'.</p>
-          </li>
-
-          <li>
-            <p>Block 'rotation.linuxnewmedia.com/'.</p>
-          </li>
-
-          <li>
-            <p>Block 'reklamy.sfd.pl/'. Reported by kacperdominik in
-            #3399948.</p>
-          </li>
-
-          <li>
-            <p>Block 'g.adspeed.net/'.</p>
-          </li>
-
-          <li>
-            <p>Unblock 'websupport.wdc.com/'. Reported by Adam Piggot in
-            #3577851.</p>
-          </li>
-
-          <li>
-            <p>Block '/openx/www/delivery/'.</p>
-          </li>
-
-          <li>
-            <p>Disable fast-redirects for '.googleapis.com/'.</p>
-          </li>
-
-          <li>
-            <p>Block 'imp.double.net/'. Reported by David Bo in #3070411.</p>
-          </li>
-
-          <li>
-            <p>Block 'gm-link.com/' whis is used for email tracking. Reported
-            by David Bo in #1812733.</p>
-          </li>
-
-          <li>
-            <p>Verify that requests to "bwp." are blocked. URL taken from
-            #1736879 submitted by Francois Marier.</p>
-          </li>
-
-          <li>
-            <p>Block '/.*bannerid='. Reported by Adam Piggott in
-            #2975779.</p>
-          </li>
-
-          <li>
-            <p>Block 'cltomedia.info/delivery/' and '.adexprt.com/'.
-            Anonymously reported in #2965254.</p>
-          </li>
-
-          <li>
-            <p>Block 'de17a.com/'. Reported by David Bo in #3061472.</p>
+            <p>Properly parse ACL directives with ports when compiled with
+            HAVE_RFC2553. Previously the port wasn't removed from the host
+            and in case of 'permit-access 127.0.0.1 example.org:80' Privoxy
+            would try (and fail) to resolve "example.org:80" instead of
+            example.org. Reported by Pak Chan on ijbswa-users@.</p>
           </li>
 
           <li>
-            <p>Block 'oskar.tradera.com/'. Reported by David Bo in
-            #3060596.</p>
+            <p>Check requests more carefully before serving them forcefully
+            when blocks aren't enforced. Privoxy always adds the force token
+            at the beginning of the path, but would previously accept it
+            anywhere in the request line. This could result in requests being
+            served that should be blocked. For example in case of pages that
+            were loaded with force and contained JavaScript to create
+            additionally requests that embed the origin URL (thus inheriting
+            the force prefix). The bug is not considered a security issue and
+            the fix does not make it harder for remote sites to intentionally
+            circumvent blocks if Privoxy isn't configured to enforce them.
+            Fixes #1695 reported by Korda.</p>
           </li>
 
           <li>
-            <p>Block '/scripts/webtrends\.js'. Reported by johnd16 in
-            #3002729.</p>
+            <p>Normalize the request line in intercepted requests to make
+            rewriting the destination more convenient. Previously rewrites
+            for intercepted requests were expected to fail unless $hostport
+            was being used, but they failed "the wrong way" and would result
+            in an out-of-memory message (vanilla host patterns) or a crash
+            (extended host patterns). Reported by "Guybrush Threepwood" in
+            #1694.</p>
           </li>
 
           <li>
-            <p>Block requests for 'pool.*.adhese.com/'. Reported by johnd16
-            in #3002716.</p>
+            <p>Enable socket lingering for the correct socket. Previously it
+            was repeatedly enabled for the listen socket instead of for the
+            accepted socket. The bug was found by code inspection and did not
+            cause any (reported) issues.</p>
           </li>
 
           <li>
-            <p>Update path pattern for Coremetrics and add tests. Pattern and
-            URLs submitted by Adam Piggott #3168443.</p>
+            <p>Detect and reject parameters for parameter-less actions.
+            Previously they were silently ignored.</p>
           </li>
 
           <li>
-            <p>Enable +fast-redirects{check-decoded-url} for 'tr.anp.se/'.
-            Reported by David Bo in #3268832.</p>
+            <p>Fixed invalid reads in internal and outdated pcre code. Found
+            with afl-fuzz and AddressSanitizer.</p>
           </li>
 
           <li>
-            <p>Unblock '.conrad.se/newsletter/banners/'. Reported by David Bo
-            in #3413824.</p>
+            <p>Prevent invalid read when loading invalid action files. Found
+            with afl-fuzz and AddressSanitizer.</p>
           </li>
 
           <li>
-            <p>Block '.tynt.com/'. Reported by Dan Stahlke in #3421767.</p>
+            <p>Windows build: Use the correct function to close the event
+            handle. It's unclear if this bug had a negative impact on
+            Privoxy's behaviour. Reported by Jarry Xu in #891.</p>
           </li>
 
           <li>
-            <p>Unblock '.bbci.co.uk/radio/'. Reported by Adam Piggott in
-            #3569603.</p>
-          </li>
-
-          <li>
-            <p>Block requests to 'service.maxymiser.net/'. Reported by
-            johnd16 in #3118401 (with a previous URL).</p>
-          </li>
-
-          <li>
-            <p>Disable fast-redirects for Google's "let's pretend your
-            computer is infected" page.</p>
-          </li>
-
-          <li>
-            <p>Unblock '/.*download' to resolve actionsfile feedback
-            #3498129. Submitted by Steven Kolins (soundcloud.com not
-            working).</p>
-          </li>
-
-          <li>
-            <p>Unblock '.wlxrs.com/' which is required by hotmail.com. Fixes
-            #3413827 submitted by David Bo.</p>
-          </li>
-
-          <li>
-            <p>Add two unblock patterns for popup radio and TV players.
-            Submitted by Adam Piggott in #3596089.</p>
+            <p>In case of invalid forward-socks5(t) directives, use the
+            correct directive name in the error messages. Previously they
+            referred to forward-socks4t failures. Reported by Joel Verhagen
+            in #889.</p>
           </li>
         </ul>
       </li>
 
       <li>
-        <p>Filter file improvements &amp; bug fixes:</p>
-
-        <ul>
-          <li>
-            <p>Add a referer tagger.</p>
-          </li>
-
-          <li>
-            <p>Reduce the likelihood that the google filter messes up
-            HTML-generating JavaScript. Reported by Zeno Kugy in
-            #3520260.</p>
-          </li>
-        </ul>
-      </li>
-
-      <li>
-        <p>Documentation improvements:</p>
-
-        <ul>
-          <li>
-            <p>Revised all OS X sections due to new packaging module
-            (OSXPackageBuilder).</p>
-          </li>
-
-          <li>
-            <p>Update the list of supported operating systems to clarify that
-            all Windows versions after 95 are expected to work and note that
-            the platform-specific code for AmigaOS and QNX currently isn't
-            maintained.</p>
-          </li>
-
-          <li>
-            <p>Update 'Signals' section, the only explicitly handled signals
-            are SIGINT, SIGTERM and SIGHUP.</p>
-          </li>
-
-          <li>
-            <p>Add Haiku to the list of operating systems on which Privoxy is
-            known to run.</p>
-          </li>
-
-          <li>
-            <p>Add DragonFly to the list of BSDs on which Privoxy is known to
-            run.</p>
-          </li>
-
-          <li>
-            <p>Removed references to redhat-specific documentation set since
-            it no longer exists.</p>
-          </li>
-
-          <li>
-            <p>Removed references to building PDFs since we no longer do
-            so.</p>
-          </li>
-
-          <li>
-            <p>Multiple listen-address directives are supported since 3.0.18,
-            correct the documentation to say so.</p>
-          </li>
-
-          <li>
-            <p>Remove bogus section about long and short being preferable to
-            int.</p>
-          </li>
-
-          <li>
-            <p>Corrected some Internet JunkBuster references to Privoxy.</p>
-          </li>
-
-          <li>
-            <p>Removed references to www.junkbusters.com since it is no
-            longer maintained. Reported by Angelina Matson.</p>
-          </li>
-
-          <li>
-            <p>Various grammar and spelling corrections</p>
-          </li>
-
-          <li>
-            <p>Add a client-header-tagger{} example for disabling filtering
-            for range requests.</p>
-          </li>
-
-          <li>
-            <p>Correct a URL in the "Privoxy with Tor" FAQ.</p>
-          </li>
-
-          <li>
-            <p>Spell 'refresh-tags' correctly. Reported by Don in
-            #3571927.</p>
-          </li>
-
-          <li>
-            <p>Sort manpage options alphabetically.</p>
-          </li>
-
-          <li>
-            <p>Remove an incorrect sentence in the toggle section. The toggle
-            state doesn't affect whether or not the Windows version uses the
-            tray icon. Reported by Zeno Kugy in #3596395.</p>
-          </li>
-
-          <li>
-            <p>Add new contributors since 3.0.19.</p>
-          </li>
-        </ul>
-      </li>
-
-      <li>
-        <p>Log message improvements:</p>
-
-        <ul>
-          <li>
-            <p>When stopping to watch a client socket due to pipelining,
-            additionally log the socket number.</p>
-          </li>
-
-          <li>
-            <p>Log the client socket and its condition before closing it.
-            This makes it more obvious that the socket actually gets closed
-            and should help when diagnosing problems like #3464439.</p>
-          </li>
-
-          <li>
-            <p>In case of SOCKS5 failures, do not explicitly log the server's
-            response. It hasn't helped so far and the response can already be
-            logged by enabling "debug 32768" anyway. This reverts v1.81 and
-            the follow-up bug fix v1.84.</p>
-          </li>
-
-          <li>
-            <p>Relocate the connection-accepted message from listen_loop() to
-            serve(). This way it's printed by the thread that is actually
-            serving the connection which is nice when grepping for thread ids
-            in log files.</p>
-          </li>
-        </ul>
-      </li>
-
-      <li>
-        <p>Code cleanups:</p>
-
-        <ul>
-          <li>
-            <p>Remove compatibility layer for versions prior to 3.0 since it
-            has been obsolete for more than 10 years now.</p>
-          </li>
-
-          <li>
-            <p>Remove the ijb_isupper() and ijb_tolower() macros from
-            parsers.c since they aren't used in this file.</p>
-          </li>
-
-          <li>
-            <p>Removed the 'Functions declared include:' comment sections
-            since they tend to be incomplete, incorrect and out of date and
-            the benefit seems questionable.</p>
-          </li>
-
-          <li>
-            <p>Various comment grammar and comprehensibility
-            improvements.</p>
-          </li>
-
-          <li>
-            <p>Remove a pointless fflush() call in chat(). Flushing all
-            streams pretty much all the time for no obvious reason is
-            ridiculous.</p>
-          </li>
-
-          <li>
-            <p>Relocate ijb_isupper()'s definition to project.h and get the
-            ijb_tolower() definition from there, too.</p>
-          </li>
-
-          <li>
-            <p>Relocate ijb_isdigit()'s definition to project.h.</p>
-          </li>
-
-          <li>
-            <p>Rename ijb_foo macros to privoxy_foo.</p>
-          </li>
-
-          <li>
-            <p>Add malloc_or_die() which will allow to simplify code paths
-            where malloc() failures don't need to be handled gracefully.</p>
-          </li>
-
-          <li>
-            <p>Add strdup_or_die() which will allow to simplify code paths
-            where strdup() failures don't need to be handled gracefully.</p>
-          </li>
-
-          <li>
-            <p>Replace strdup() calls with strdup_or_die() calls where it's
-            safe and simplifies the code.</p>
-          </li>
-
-          <li>
-            <p>Fix white-space around parentheses.</p>
-          </li>
-
-          <li>
-            <p>Add missing white-space behind if's and the following
-            parentheses.</p>
-          </li>
-
-          <li>
-            <p>Unwrap a memcpy() call in resolve_hostname_to_ip().</p>
-          </li>
-
-          <li>
-            <p>Declare pcrs_get_delimiter()'s delimiters[] static const.</p>
-          </li>
-
-          <li>
-            <p>Various optimisations to remove dead code and merge
-            inefficient code structures for improved clarity, performance or
-            code compactness.</p>
-          </li>
-
-          <li>
-            <p>Various data type corrections.</p>
-          </li>
-
-          <li>
-            <p>Change visibility of several code segments when compiling
-            without FEATURE_CONNECTION_KEEP_ALIVE enabled for clarity.</p>
-          </li>
-
-          <li>
-            <p>In pcrs_get_delimiter(), do not use delimiters ouside the
-            ASCII range. Fixes a clang complaint.</p>
-          </li>
-
-          <li>
-            <p>Fix an error message in get_last_url() nobody is supposed to
-            see. Reported by Matthew Fischer in #3507301.</p>
-          </li>
-
-          <li>
-            <p>Fix a typo in the no-zlib-support complaint. Patch submitted
-            by Matthew Fischer in #3507304.</p>
-          </li>
-
-          <li>
-            <p>Shorten ssplit()'s prototype by removing the last two
-            arguments. We always want to skip empty fields and ignore leading
-            delimiters, so having parameters for this only complicates the
-            API.</p>
-          </li>
-
-          <li>
-            <p>Use an enum for the type of the action value.</p>
-          </li>
-
-          <li>
-            <p>Rename action_name's member takes_value to value_type as it
-            isn't used as boolean.</p>
-          </li>
-
-          <li>
-            <p>Turn family mismatches in match_sockaddr() into fatal
-            errors.</p>
-          </li>
-
-          <li>
-            <p>Let enlist_unique_header() verify that the caller didn't pass
-            a header containing either \r or \n.</p>
-          </li>
-
-          <li>
-            <p>Change the hashes used in load_config() to unsigned int.
-            That's what hash_string() actually returns and using a
-            potentiallly larger type is at best useless.</p>
-          </li>
-
-          <li>
-            <p>Use privoxy_tolower() instead of vanilla tolower() with manual
-            casting of the argument.</p>
-          </li>
-
-          <li>
-            <p>Catch ssplit() failures in parse_cgi_parameters().</p>
-          </li>
-        </ul>
-      </li>
-
-      <li>
-        <p>Privoxy-Regression-Test:</p>
+        <p>General improvements:</p>
 
         <ul>
           <li>
-            <p>Add an 'Overwrite condition' directive to skip any matching
-            tests before it. As it has a global scope, using it is more
-            convenient than clowning around with the Ignore directive.</p>
+            <p>Set NO_DELAY flag for the accepting socket. This significantly
+            reduces the latency if the operating system is not configured to
+            set the flag by default. Reported by Johan Sintorn in #894.</p>
           </li>
 
           <li>
-            <p>Log to STDOUT instead of STDERR.</p>
+            <p>Allow to build with mingw x86_64. Submitted by Rustam
+            Abdullaev in #135.</p>
           </li>
 
           <li>
-            <p>Include the Privoxy version in the output.</p>
+            <p>Introduce the new forwarding type 'forward-webserver'.
+            Currently it is only supported by the forward-override{} action
+            and there's no config directive with the same name. The
+            forwarding type is similar to 'forward', but the request line
+            only contains the path instead of the complete URL.</p>
           </li>
 
           <li>
-            <p>Various grammar and spelling corrections in documentation and
-            code.</p>
+            <p>The CGI editor no longer treats 'standard.action' special.
+            Nowadays the official "standards" are part of default.action and
+            there's no obvious reason to disallow editing them through the
+            cgi editor anyway (if the user decided that the lack of
+            authentication isn't an issue in her environment).</p>
           </li>
 
           <li>
-            <p>Additional tests for range requests with filtering
-            enabled.</p>
+            <p>Improved error messages when rejecting intercepted requests
+            with unknown destination.</p>
           </li>
 
           <li>
-            <p>Tests with mostly invalid range request.</p>
+            <p>A couple of log messages now include the number of active
+            threads.</p>
           </li>
 
           <li>
-            <p>Add a couple of hide-if-modified-since{} tests with different
-            date formats.</p>
+            <p>Removed non-standard Proxy-Agent headers in HTTP snipplets to
+            make testing more convenient.</p>
           </li>
 
           <li>
-            <p>Cleaned up the format of the regression-tests.action file to
-            match the format of default.action.</p>
+            <p>Include the error code for pcre errors Privoxy does not
+            recognize.</p>
           </li>
 
           <li>
-            <p>Remove the "Copyright" line from print_version(). When using
-            --help, every line of screen space matters and thus shouldn't be
-            wasted on things the user doesn't care about.</p>
-          </li>
-        </ul>
-      </li>
-
-      <li>
-        <p>Privoxy-Log-Parser:</p>
-
-        <ul>
-          <li>
-            <p>Improve the --statistics performance by skipping sanity checks
-            for input that shouldn't affect the results anyway. Add a
-            --strict-checks option that enables some of the checks again,
-            just in case anybody cares.</p>
+            <p>Config directives with numerical arguments are checked more
+            carefully.</p>
           </li>
 
           <li>
-            <p>The distribution of client requests per connection is included
-            in the --statistic output.</p>
+            <p>Privoxy's malloc() wrapper has been changed to prevent
+            zero-size allocations which should only occur as the result of
+            bugs.</p>
           </li>
 
           <li>
-            <p>The --accept-unknown-messages option has been removed and the
-            behavior is now the default.</p>
-          </li>
-
-          <li>
-            <p>Accept and (mostly) highlight new log messages introduced with
-            Privoxy 3.0.20.</p>
-          </li>
-        </ul>
-      </li>
-
-      <li>
-        <p>uagen:</p>
-
-        <ul>
-          <li>
-            <p>Bump generated Firefox version to 17.</p>
+            <p>Various cosmetic changes.</p>
           </li>
         </ul>
       </li>
 
       <li>
-        <p>GNUmakefile improvements:</p>
+        <p>Action file improvements:</p>
 
         <ul>
           <li>
-            <p>The dok-tidy target no longer taints documents with a
-            tidy-mark</p>
+            <p>Unblock ".deutschlandradiokultur.de/". Reported by u302320 in
+            #924.</p>
           </li>
 
           <li>
-            <p>Change RA_MODE from 0664 to 0644. Suggested by Markus Dittrich
-            in #3505445.</p>
+            <p>Add two fast-redirect exceptions for "yandex.ru".</p>
           </li>
 
           <li>
-            <p>Remove tidy's clean flag as it changes the scope of
-            attributes. Link-specific colors end up being applied to all
-            text. Reported by Adam Piggott in #3569551.</p>
+            <p>Disable filter{banners-by-size} for ".plasmaservice.de/".</p>
           </li>
 
           <li>
-            <p>Leave it up to the user whether or not smart tags are
-            inserted.</p>
+            <p>Unblock "klikki.fi/adv/".</p>
           </li>
 
           <li>
-            <p>Let w3m itself do the line wrapping for the config file. It
-            works better than fmt as it can honour pre tags causing less
-            unintentional line breaks.</p>
+            <p>Block requests for "resources.infolinks.com/". Reported by
+            "Black Rider" on ijbswa-users@.</p>
           </li>
 
           <li>
-            <p>Ditch a pointless '-r' passed to rm to delete files.</p>
+            <p>Block a bunch of criteo domains. Reported by Black Rider.</p>
           </li>
 
           <li>
-            <p>The config-file target now requires less manual intervention
-            and updates the original config.</p>
+            <p>Block "abs.proxistore.com/abe/". Reported by Black Rider.</p>
           </li>
 
           <li>
-            <p>Change WDUMP to generate ASCII. Add WDUMP_UTF8 to allow UTF-8
-            in the AUTHORS file so the names are right.</p>
+            <p>Disable filter{banners-by-size} for
+            ".black-mosquito.org/".</p>
           </li>
 
           <li>
-            <p>Stop pretending that lynx and links are supported for the
-            documentation.</p>
+            <p>Disable fast-redirects for "disqus.com/".</p>
           </li>
         </ul>
       </li>
 
       <li>
-        <p>configure improvements:</p>
+        <p>Documentation improvements:</p>
 
         <ul>
           <li>
-            <p>On Haiku, do not pass -lpthread to the compiler. Haiku's
-            pthreads implementation is contained in its system library,
-            libroot, so no additional library needs to be searched. Patch
-            submitted by Simon South in #3564815.</p>
+            <p>FAQ: Explicitly point fingers at ASUS as an example of a
+            company that has been reported to force malware based on Privoxy
+            upon its customers.</p>
           </li>
 
           <li>
-            <p>Additional Haiku-specific improvements. Disable checks
-            intended for multi-user systems as Haiku is presently
-            single-user. Group Haiku-specific settings in their own section,
-            following the pattern for Solaris, OS/2 and AmigaOS. Add
-            additional library-related settings to remove the need for
-            providing configure with custom LDFLAGS. Submitted by Simon South
-            in #3574538. *** Version 3.0.19 Stable ***</p>
-          </li>
-        </ul>
-      </li>
-
-      <li>
-        <p>Bug fixes:</p>
-
-        <ul>
-          <li>
-            <p>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.</p>
+            <p>Correctly document the action type for a bunch of
+            "multi-value" actions that were incorrectly documented to be
+            "parameterized". Reported by Gregory Seidman on
+            ijbswa-users@.</p>
           </li>
 
           <li>
-            <p>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.</p>
+            <p>Fixed the documented type of the forward-override{} action
+            which is obviously 'parameterized'.</p>
           </li>
         </ul>
       </li>
 
       <li>
-        <p>General improvements:</p>
+        <p>Website improvements:</p>
 
         <ul>
           <li>
-            <p>Fix an off-by-one in an error message about connect
-            failures.</p>
+            <p>Users who don't trust binaries served by SourceForge can get
+            them from a mirror. Migrating away from SourceForge is planned
+            for 2016 (TODO list item #53).</p>
           </li>
 
           <li>
-            <p>Use a GNUMakefile variable for the webserver root directory
-            and update the path. Sourceforge changed it which broke various
-            web-related targets.</p>
-          </li>
-
-          <li>
-            <p>Update the CODE_STATUS description.</p>
+            <p>The website is now available as onion service
+            (http://jvauzb4sb3bwlsnc.onion/).</p>
           </li>
         </ul>
       </li>
@@ -945,11 +347,6 @@
           yourself.</p>
         </li>
 
-        <li>
-          <p><tt class="FILENAME">standard.action</tt> has been merged into
-          the <tt class="FILENAME">default.action</tt> file.</p>
-        </li>
-
         <li>
           <p>In the default configuration only fatal errors are logged now.
           You can change that in the <a href="config.html#DEBUG">debug