This file belongs into
ijbswa.sourceforge.net:/home/groups/i/ij/ijbswa/htdocs/
- $Id: user-manual.sgml,v 2.100 2009/02/19 17:14:11 fabiankeil Exp $
+ $Id: user-manual.sgml,v 2.101 2009/02/25 19:01:56 fabiankeil Exp $
Copyright (C) 2001-2009 Privoxy Developers http://www.privoxy.org/
See LICENSE.
</subscript>
</pubdate>
-<pubdate>$Id: user-manual.sgml,v 2.100 2009/02/19 17:14:11 fabiankeil Exp $</pubdate>
+<pubdate>$Id: user-manual.sgml,v 2.101 2009/02/25 19:01:56 fabiankeil Exp $</pubdate>
<!--
<sect1 id="whatsnew">
<title>What's New in this Release</title>
<para>
- There are only a few improvements and new features since
- <application>Privoxy 3.0.10</application>, the last stable release:
+ <application>Privoxy 3.0.11</application> is mainly a bugfix release:
</para>
<para>
<itemizedlist>
<listitem>
<para>
- On most platforms, outgoing connections can be kept alive and
- reused if the server supports it. Whether or not this improves
- things depends on the connection.
+ The socket-timeout option now also works on platforms whose
+ select() implementation modifies the timeout structure.
+ Previously the timeout was triggered even if the connection
+ didn't stall. Reported by cyberpatrol.
</para>
</listitem>
<listitem>
<para>
- When dropping privileges, membership in supplementary groups
- is given up as well. Not doing that can lead to Privoxy running
- with more rights than necessary and violates the principle of
- least privilege. Users of the --user option are advised to update.
- Thanks to Matthias Drochner for reporting the problem,
- providing the initial patch and testing the final version.
+ The Connection: keep-alive code properly deals with files
+ larger than 2GB. Previously the connection was closed too
+ early.
</para>
</listitem>
<listitem>
<para>
- Passing invalid users or groups with the --user option
- didn't lead to program exit. Regression introduced in 3.0.7.
+ The content length for files above 2GB is logged correctly.
</para>
</listitem>
<listitem>
<para>
- The match all section has been moved from default.action
- to a new file called match-all.action. As a result the
- default.action no longer needs to be touched by the user
- and can be safely overwritten by updates.
+ The user-manual directive on the show-status page links to
+ the documentation location specified with the directive,
+ not to the Privoxy website.
</para>
</listitem>
<listitem>
<para>
- The standard.action file has been removed. Its content
- is now part of the default.action file.
+ The show-status page prints warnings about invalid directives
+ on the same line as the directives themselves.
</para>
</listitem>
<listitem>
<para>
- In some situations the logged content length was slightly too low.
+ Fixed several justified (but harmless) compiler warnings,
+ mostly on 64 bit platforms.
</para>
</listitem>
<listitem>
<para>
- Crunched requests are logged with their own log level.
- If you used "debug 1" in the past, you'll probably want
- to additionally enable "debug 1024", otherwise only passed
- requests will be logged. If you only care about crunched
- requests, simply replace "debug 1" with "debug 1024".
+ The mingw32 version explicitly requests the default charset
+ to prevent display problems with some fonts available on more
+ recent Windows versions. Patch by Burberry.
</para>
</listitem>
<listitem>
<para>
- The crunch reason has been moved to the beginning of the
- crunch message. For HTTP URLs, the protocol is logged as well.
+ The mingw32 version uses the Privoxy icon in the alt-tab
+ windows. Patch by Burberry.
</para>
</listitem>
<listitem>
<para>
- Log messages are shortened by printing the thread id on its
- own (as opposed to putting it inside the string "Privoxy()").
+ The timestamp and the thread id is omitted in the "Fatal error"
+ message box on mingw32.
</para>
</listitem>
<listitem>
<para>
- The config option socket-timeout has been added to control
- the time Privoxy waits for data to arrive on a socket.
+ Fixed two related mingw32-only buffer overflows. Triggering
+ them required control over the configuration file, therefore
+ this isn't seen as a security issue.
</para>
</listitem>
<listitem>
<para>
- Support for remote toggling is controlled by the configure
- option --disable-toggle only. In previous versions it also
- depended on the action editor and thus configuring with the
- --disable-editor option would disable remote toggling support
- as well.
- </para>
- </listitem>
- <listitem>
- <para>
- Requests with invalid HTTP versions are rejected.
- </para>
- </listitem>
- <listitem>
- <para>
- The template symbol @date@ can be used to include a date(1)-like
- time string. Initial patch submitted by Endre Szabo.
- </para>
- </listitem>
- <listitem>
- <para>
- Responses from shoutcast servers are accepted again.
- Problem reported and fix suggested by Stefan.
- </para>
- </listitem>
- <listitem>
- <para>
- The hide-forwarded-for-headers action has been replaced with
- the change-x-forwarded-for{} action which can also be used to
- add X-Forwarded-For headers. The latter functionality already
- existed in Privoxy versions prior to 3.0.7 but has been removed
- as it was often used unintentionally (by not using the
- hide-forwarded-for-headers action).
- </para>
- </listitem>
- <listitem>
- <para>
- A "clear log" view option was added to the mingw32 version
- to clear out all of the lines in the Privoxy log window.
- Based on a patch submitted by T Ford.
- </para>
- </listitem>
- <listitem>
- <para>
- The mingw32 version uses "critical sections" now, which prevents
- log message corruption under load. As a side effect, the
- "no thread-safe PRNG" warning could be removed as well.
- </para>
- </listitem>
- <listitem>
- <para>
- The mingw32 version's task bar icon is crossed out and
- the color changed to gray if Privoxy is toggled off.
+ In verbose mode, or if the new option --show-skipped-tests
+ is used, Privoxy-Regression-Test logs skipped tests and the
+ skip reason.
</para>
</listitem>
</itemizedlist>
</para>
-<para>
- This release marks a departure for Privoxy development.
-</para>
-<para>
- Previously, odd numbered releases were considered beta versions and
- were only released at the end of the development cycle when the code
- was already believed to be stable. Usually it was, so the stable release
- contained pretty much the same code, but got a higher version number.
- In the future we intend to release several snapshots between stable releases.
- There will probably still be about two stable releases per year,
- but hopefully about six snapshots instead of the two betas we have now.
- The intentions is to make testing without CVS access easier.
-</para>
<!-- ~~~~~ New section ~~~~~ -->
USA
$Log: user-manual.sgml,v $
+ Revision 2.101 2009/02/25 19:01:56 fabiankeil
+ Fix typo.
+
Revision 2.100 2009/02/19 17:14:11 fabiankeil
- Copy the release cycle description from announce.txt into
the "What's New" section.