1 $Id: TODO,v 1.150 2016/04/10 08:07:10 fabiankeil Exp $
3 Some Privoxy-related tasks, sorted by the time they
4 have been added, not by priority.
6 The latest version should be available at:
7 http://ijbswa.cvs.sourceforge.net/viewvc/ijbswa/current/TODO
9 There's work in progress to fund development on these items using
10 donations. If you want to donate, please have a look at:
11 http://www.privoxy.org/faq/general.html#DONATE
13 1) Add more regression tests. Filters should be tested automatically
14 (variables too). Could probably reuse large parts of Privoxy-Filter-Test.
15 Note that there is currently work in progress to leverage curl's
16 test suite, patches have been submitted upstream:
17 http://curl.haxx.se/mail/lib-2014-06/0070.html
19 3) Fix some more XXX: comments.
21 6) Remove actions that aren't needed anymore:
23 content-type-overwrite should probably stay as it's also
24 used by some of the CGI pages (XXX: name them).
26 crunch-client-header and crunch-server-header should probably
27 go, their only advantage is that their search strings can be
28 controlled through the CGI pages, other than that they only
31 crunch-if-none-match can be replaced with a header filter.
33 prevent-compression has a misleading name and could
34 be replaced with a header filter.
36 7) force-text-mode has a stupid name and should probably
37 be renamed to force-filter-mode.
39 8) handle-as-empty-document and handle-as-image should
40 be merged to something like handle-as{something} to
41 prevent them from being activated at the same time.
43 10) There's a bug in the CGI editor that turns the
44 first section's "Insert new section below" into
45 a "Insert new section above" button.
47 11) CGI templates should use semantically-correct HTML
52 12) Support pipelining for outgoing connections.
54 14) Allow to filter POST parameters.
56 15) If trusted CGI pages are requested without trusted referrer,
57 set the status code to 403 instead of 200.
59 16) Filter SSL encrypted content as well.
61 At the beginning we could use a unencrypted connection between
62 client and Privoxy, and use an encrypted connection between
63 Privoxy and the server.
65 This should be good enough for most of the content the
66 user would want to filter.
70 19) enable-forward-fallback. Syntax? Suggested by K.R.
72 21) User Manual delivery doesn't accept multiple slashes. Should it?
74 22) Verify action files properly (Including arguments) and
75 act accordingly (should probably intercept all requests
76 with a "Invalid option foo detected" CGI page).
78 23) Do the same in case of syntax errors in the configuration file,
79 instead of just exiting or ignoring the problem.
81 25) Handle multiple filters with the same name better. Reject them?
83 26) Let show-url-info detect clearly invalid URLs.
85 27) Make errno logging less thread-unsafe.
86 Verify that it's really an improvement.
88 28) Don't take default ports in case of invalid forwarding ports.
90 31) If a string action foo is disabled csp->action->string[ACTION_STRING_FOO]
91 doesn't necessarily contain NULL, but may contain the string of an
92 enabled foo action in an overruled section. Is it a bug? Does it matter?
94 32) In case of forwarding failures with socks port == 9050,
95 show extra info about Tor (the whole FAQ entry?).
97 36) Unload unused action files directly, even if they are
98 disabled without replacement.
100 38) In the final results, explicitly list disabled multi actions
101 with their parameters. Not as trivial as it sounds.
103 40) When running in daemon mode, Privoxy's working directory is '/'
104 which means it may not have permissions to dump core when necessary.
105 Figure out a way to solve this. Introduce a cwd config option?
107 41) Change documentation framework to one that works cross-platform.
108 Evaluate WML and txt2tags.
110 42) Add a DTrace USDT provider. Now that FreeBSD has userland DTrace
111 support there's no longer any reason not to.
113 43) Write a tool to check URL patterns against URLs in the log file.
114 This could be included in Privoxy-Regression-Test.
116 44) Privoxy-Log-Parser: Consider highlighting "Connection" in:
117 23:13:03.506 283b6100 Header: Replaced: 'Connection: Keep-Alive' with 'Connection: close'
119 50) Investigate possible PCRS template speedup when searching
120 macros with strstr() before compiling pcrs commands.
121 Investigated, needs some restructuring but is probably worth it.
123 51) Make user-manual directive more generic to allow serving the FAQ
124 and files from user-specified directories. Consider changing the
125 port for "same origin policy" issues.
127 53) Find a more reliable hoster. Involves finding out what our
128 requirements are and which SF alternatives fulfil them.
129 It would probably also make sense to look into what other
130 projects did when migrating away from SF.
132 2014-05: Work in progress. Hosting wish list at the end
133 of this file. Looks like most of the other projects
134 that left SF had lower standards and moved to hosters
135 that don't come close to sattisfying the requirements.
136 2016-03: The website has been moved away from SF infrastructure
137 and is also available through https:// now.
138 2016-04: Server rent for a year has been sponsored by ChameleonJohn.
140 Interested donors: 1.
142 54) Move away from CVS to a more modern revision control system.
143 The move to git is work in progress:
144 https://sourceforge.net/p/ijbswa/mailman/message/34994343/
146 58) Move more template strings from the code into the actual templates.
148 59) Import the German template translation.
150 60) Ask the Russian translators for input on how to make their
153 61) Consider (optionally?) skipping the hostname comparison when
154 checking if a connections that goes to a HTTP proxy can be reused.
155 Do all HTTP proxy support that? Is it worth it?
157 63) Reject clearly too large requests earlier?
159 64) Use proper copyright attribution. "Privoxy Developers"
162 65) Polish Website. Probably involves ditching the Docbook
163 mess. There are already several threads in the mailinglist
164 archives about this. See also #41.
166 66) Stop hard-coding the number of action and filter files.
168 67) Clean up source code directory layout. Depends on 54 so
169 we don't lose the revision history.
171 68) Use standard make syntax so we don't depend on GNU make.
173 69) Update autoconf setup (or move away from it).
175 70) If the server connection is reset but the headers are
176 received, consider passing the mess to the client instead
177 of showing the connect-failed template. Relates to #2698674.
179 74) Let Privoxy-Regression-Test optionally check that action
180 sections which disable actions actually are preceded by
181 sections that enable said actions.
183 75) Create a tool that creates Privoxy action (and filter?) files
184 out of adblock files. Could be implemented as option for
185 url-pattern-translator.pl.
187 76) Cache DNS responses. Note that this has been requested
188 several times by users, but is not a developer priority.
189 If you care about this, feel free to submit patches.
191 77) Allow to configure the IP address used in outgoing connections.
193 78) Allow to optionally use pcre's DFA algorithm.
195 79) Evaluate pcre alternatives.
197 80) Change FEATURE_EXTENDED_HOST_PATTERNS to support both
198 extended and vanilla host patterns at the same time.
200 Note that the requirement is to allow the user to decide
201 if the domain pattern should be interpreted as regex or
202 traditional host pattern and if it's not obvious that the
203 user made any decision, default to the latter.
205 Possible solutions would be:
207 1. An always-use-regex-domain-patterns config option
208 2. An enable-regex-domain-patterns-for-this-action-file option
209 3. An enable-regex-domain-patterns-for-this-action-file-until-the-user-says-otherwise option
210 4. A treat-the-domain-pattern-in-this-line-as-regex(-or-not) option
211 5. Combinations of the options above
213 With 2+4, 3+4 or 2+3+4 being the preferences until
216 82) Detect if the system time goes back in time let the user
217 know if it caused any connections to get closed.
219 85) Once #84 is done, write a script that populates a directory with
220 various common third-party icons (stumbleupon.png, facebook.png ...)
221 and redirect requests for them to Privoxy.
223 86) Add a server-body-tagger action. This is trivial as as all the
224 functionality required to do it already exists.
226 87) Add a client-body-tagger action. This is less trivial as we currently
227 don't buffer client bodies. After 14) is implemented it would be
230 88) Investigate if there's a Perl module that Privoxy-Regression-Test
231 could optionally use to keep connections alive, preferably while
232 requiring less forks at the same time.
234 89) When multiple block actions apply, consider showing all the block
235 reasons on the blocked page that haven't been overruled, not just
238 91) Add an optional limit for internal redirects. It would probably
239 be reasonable to default to a limit of one and showing an error
240 message if the request for the redirect URL would be redirected
243 92) The statistics currently aren't calculated correctly by Privoxy
244 as each thread is only counted as one request which is no longer
245 correct. This should be fixed, or the statistic code removed.
246 Privoxy-Log-Parser's provides more detailed statistics, anyway.
248 93) Add a config directive to let Privoxy explicitly request either
249 IPv4 (or IPv6) addresses, even if the system supports both.
250 Could be useful as a workaround for misconfigured setups where the
251 libc returns IPv6 addresses even if there's no IPv6 connectivity.
253 94) Add a config directive to let Privoxy prefer either IPv4 (or IPv6)
254 addresses, instead of trusting the libc to return them in an order
255 that makes sense. Like #93, this could be useful as a workaround
256 for misconfigured setups.
258 95) Support a non-standard client header in CONNECT requests that
259 contains the URL of the requested resource, which is then treated
260 like the request URL.
262 This way the client could opt-in for path-based blocking of https
263 requests. Given that the headers from the CONNECT request aren't
264 forwarded to the destination server, an unencrypted URL should be
265 acceptable if the client and Privoxy are running on the same system
266 or in a trusted environment.
268 96) Filters should be easier to look up. Currently get_filter() has to
269 go through all filters and skip the filter types the caller isn't
272 98) When showing action section on the CGI pages, properly escape
273 line breaks so they can be copy&pasted into action files without
276 99) Figure out a mechanism through which a user can easily enable
277 site-specific action sections that are too aggressive to be
278 enabled by default. This could be similar to the presettings
279 in default.action, but could also be just another action file
280 that isn't used by default.
282 100) Create a cross-platform Privoxy control program and retire
283 the win32 GUI. Integrate support for Privoxy-Regression-Test,
284 Privoxy-Log-Parser, Privoxy-Filter-Test, uagen and similar tools.
285 Interested donors: 1.
287 102) Add an include directive to split the config file into several parts.
289 103) Potential performance improvement for large action files:
290 when figuring out which actions apply, check the action bit mask
291 before pattern matching and skip section that wouldn't modify the
292 actions already set. To increase the impact the sections would have
293 to be applied in reverse.
295 104) The code to modify global_toggle_state should be factored out into
296 a separate function. Currently we mess with it in three different
297 files, but only in w32log.c the tray icon is explicitly set.
298 The logging is inconsistent as well. For details see #3525694.
300 105) Add support for socks authentication.
302 106) actionlist.h should be embedded in a way that causes less text
305 107) Support more pcrs variables, for example $destination-ip-address
306 and $source-ip-address.
308 108) Allow to use a somewhat random string instead of PRIVOXY-FORCE.
310 109) Let log_error() support the format specifier %S which should
311 work like %s but escape new lines like %N. This would be useful
312 to log the result of header filters which may inject new lines.
314 110) Add a global-buffer-limit directive that roughly limits how
315 much malloc'ed memory Privoxy will use and can potentially
316 be smaller than (buffer-limit * max-client-connections).
318 111) Reject requests if hosts and ports in request line and Host
319 header don't match (before filters have been applied).
321 112) If a header filter is used to inject another header by inserting
322 a \r\n (undocumented feature), detect it and split the headers so
323 following header actions do not treat them as a single string.
324 Alternatively add another header injection mechanism.
326 113) Log statistics upon receiving a certain signal (SIGINFO or SIGUSR1).
328 114) Properly deal with status code 100. The current "Continue hack"
329 can cause problems for gpg when uploading keys through Privoxy.
331 115) Add ICAP (RFC 3507) support. FR #3615158.
333 116) Due to the use of sscanf(), Privoxy currently will fail to properly
334 parse chunks whose size can't be represented with 32 bit. This is
335 unlikely to cause problems in the real world, but should eventually
336 be fixed anyway. See also:
337 https://bugzilla.mozilla.org/show_bug.cgi?id=959100
339 118) There should be "escaped" dynamic variables that are guaranteed
340 not to break filters.
342 119) Evaluate using pcre's jit mode.
344 120) Add an option to limit pcre's recursion limit below the default.
345 On some platforms the recursion limit doesn't prevent pcre from
346 running out of stack space, causing the kernel to kill Privoxy
349 121) Add HTTP/2 support. As a first step, incomming HTTP/1.x requests
350 should be translated to outgoing HTTP/2 requests where possible
351 (and if desired by the user).
352 Interested donors: 1.
354 122) Allow customized log messages.
356 123) Evaluate if the voluntarily-disclose-session-keys option in Firefox
357 (and other browsers) can be leveraged. Probably depends on #16.
359 124) Add support for the "lightweight OS capability and sandbox framework"
360 Capsicum. http://www.cl.cam.ac.uk/research/security/capsicum/
361 Interested donors: 1.
363 125) Allow clients to HTTPS-encrypt the proxy connection.
364 Interested donors: 1.
366 126) Run the Co-Advisor HTTP compliance tests, evaluate the results,
367 fix the compliance issues that aren't by design and document
369 Note that Privoxy developers qualify for free account upgrades:
370 http://coad.measurement-factory.com/details.html#pricing
372 127) Add "real" CGI support (serve program output instead of forwarding
373 the request). The work is mostly done due to +external-filter{}.
375 128) Add a config directive to control the stack limit.
377 129) Completely implement RFC 7230 4.1 (Chunked Transfer Coding).
378 Currently Privoxy doesn't properly deal with trailers which
379 are rarely used in the real world but should be supported anyway.
381 130) Move header_tagger() out of the parser structs and let it execute
382 taggers one-by-one against all headers so the header order has less
383 influence on the tagging result. As a bonus, dynamic taggers would
384 have to be compiled less often.
386 131) The handle-as-empty-doc-returns-ok directive should be replaced with
387 an action so the behaviour can be enabled on a per-request basis.
388 Interested donors: 1.
390 133) Consider allowing bitcoin donations.
391 Interested donors: 2.
393 134) Track the total number of bytes written to and received from a socket.
395 135) Add OpenBSM audit support.
397 136) Make builds reproducible.
399 137) Add a (preferably vector-based) logo.
401 138) Bring back the scripts to provide actions file feedback.
403 Once upon a time (~2003) there were scripts on the webserver
404 to make reporting action file feedback more convenient for the
405 user and the actual reports more useful for the developers.
406 They have been unusable for years and have thus been disabled,
407 but making the reporting mechanism available again would be a
410 140) Toggling Privoxy off currently also disables stuff that
411 probably shouldn't be affected (such as actions like
412 forward-override). Investigate and fix or document.
414 141) Port Privoxy to CloudABI, which, despite the name, is actually
415 rather neet. https://github.com/NuxiNL/cloudlibc
417 142) Remove or update the "internal" pcre version.
419 143) Add support for OpenBSD's pledge feature once it's stablelized.
420 This should be a lot less work then #124.
422 146) Allow to save the internal client tag state to disk and
423 load it after restarts.
425 147) Improve "Building from Source" section in the user manual.
426 A common problem seems to be that it's not obvious to non-technical
427 users how the listed dependencies can be installed on the commonly
428 used platforms. Adding a couple of examples should also be useful for
429 technical users (like Privoxy developers) who want to install or test
430 Privoxy on platforms they are not familiar with.
432 148) Add a config directive to change the CGI_SITE_2_HOST
433 (default: config.privoxy.org).
435 If Privoxy is used as reverse proxy or intercepting proxy without
436 getting intercepted requests, error pages created from default templates
437 currently can result in client requests to config.privoxy.org on the
438 Internet which may not be desirable.
440 149) Use poll() for socket selection so the number of sockets Privoxy
441 can deal with isn't limited to FD_SETSIZE anymore.
443 150) Add blacklistd support.
445 ##########################################################################
447 Hosting wish list (relevant for #53)
452 - Mailinglists (Mailman with public archives preferred)
453 - Webspace (on a Unix-like OS that works with the webserver targets
455 - Source code repositories (currently CVS, but migrating away
456 from it is TODO #54 anyway and shouldn't be too much trouble)
457 - Commit mails (preferably with unified diffs)
459 (Unsorted) details to look at when evaluating hosters:
461 1. Preferably no third-party ads and trackers.
462 External images, CSS and JavaScript may count as trackers
463 but texts like "supported by company XYZ" may be acceptable.
465 2. JavaScript should be optional or not used at all.
467 3. Services we don't need shouldn't be enabled anyway.
468 (We currently don't use Web forums, wikis, surveys etc.)
470 4. It would be preferable if the hoster didn't have a bad track
471 record as far as user experience, security and privacy are
472 concerned and if the terms of service are "reasonable" and
473 haven't changed too often in the past. Updates in the past
474 should have been improvements and not regressions.
476 5. It would be preferable if most of the server administration
477 is done by a trusted third-party (or at least not a lot of work
480 6. The server(s) should be located in a country with laws we can
481 understand and follow (or at least not unintentionally violate).
483 7. A server location in a country with some kind of due process
484 and strong data protection laws (at least on paper) would be
487 8. Given that Privoxy is a free software project it would be
488 preferable if the hoster would use free software where possible.
490 9. Migrating away from the hoster in the future without losing
491 any important data should be possible without writing web